Arakawa, Hiroshi (Sagamihara, JP)
Ninose, Kenta (Yokohama, JP)
Deguchi, Akira (Yokohama, JP)
Homma, Hisao (Odawara, JP)
Okumoto, Katsuhiro (Odawara, JP)

11/251212

09/22/2009

10/17/2005

Download PDF 7594086       PDF help

Click for automatic bibliography generation

Hitachi, Ltd. (Tokyo, JP)

711/203

711/161, 711/162

G06F12/00; G06F13/00; G06F13/28

711/203

20060212669	Control method for storage system, storage system, storage control apparatus, control program for storage system, and information processing system	September, 2006	Uchida et al.	711/162
20060224845	Computer system for remote copy	October, 2006	Hiraiwa et al.	711/162
20060236047	Method for replicating snapshot volumes between storage systems	October, 2006	Shitomi	711/162
20060236048	Method for remote copy pair migration	October, 2006	Deguchi et al.	711/162
20060236049	Remote copy system and remote copy method	October, 2006	Iwamura	711/162
20060259722	Remote copy system	November, 2006	Watanabe	711/162
20080140966	Method for remote copy pair migration	June, 2008	Deguchi et al.	711/162

JP2000242437	September, 2000			STORAGE DEVICE SYSTEM FOR PREPARING COPY OF DATA
JP20045370	January, 2004
JP2005078453	March, 2005			METHOD FOR CONTROLLING STORAGE SYSTEM, STORAGE SYSTEM, AND STORAGE DEVICE

“Using Asynchronous Replication for Business Continuity between Two or More Sites” Dec. 2004. pp. 1-20.

Kim, Matt

Chrzanowski, Matthew R.

Brundidge & Stanger, P.C.

What is claimed is:

1. A storage system for providing a host computer with a data storage area, comprising: a primary storage device connected to the host computer and having a primary storage area for the host computer; a first secondary storage device connected to the primary storage device and having a first secondary storage area; and a second secondary storage device connected to the primary storage device and having a second secondary storage area, the primary storage device comprising: a virtualization unit configured to set a correlation between the second secondary storage area and a virtual identifier used to identify the second secondary storage area as a virtual storage area within the primary storage device; a primary writing unit configured to execute a primary write process, the primary write process including a process of receiving from the host computer a host request which is a data write request of host data, and a process of writing the host data to the primary storage area; and a copy control unit configured to send a first and second copy request to the first and second secondary storage devices, respectively, to store copy data in the first and second secondary storage areas, the copy data being a copy of the host data, the first and second copy request containing the copy data, wherein the copy control unit, regardless of whether or not receiving a completion notification of the second copy request from the second secondary storage device, sends a completion notification of the host request to the host computer responsive to receiving a completion notification of the first copy request from the first secondary storage device, and the copy control unit identifies the second secondary storage area using the virtual identifier in the second copy request; wherein the copy control unit comprises: a copy instruction unit configured to create the first and second copy request, the copy instruction unit creating the second copy request identifying the second secondary storage area using the virtual identifier; and an access control unit configured to send the first and second copy requests to the first and second secondary storage devices, respectively, wherein, the access control unit, by referencing the correlation set by the virtualization unit, replaces the virtual identifier within the second copy request with an actual identifier of the second secondary storage area, and sends the second copy request after the replacement to the second secondary storage device.

2. A storage system recited in claim 1, wherein the first secondary storage device and the second secondary storage device are connected to each other, and the primary storage device further comprises: a generation setting unit configured to allocate a generation to the host request and to update the generation according to specified conditions, the generation representing a time range in which the primary write process is executed; and a generation information sending unit configured to send first generation information to the first secondary storage device, the first generation information representing the generation of the host request that is a source of the first copy request, the first secondary storage device further comprises: a first history creation unit configured to create first history information for each generation of host requests which are sources of the first copy requests by using the first generation information, the first history information representing a write history according to the first copy requests; and a secondary copy unit configured to execute a secondary copy process of copying data between a storage area within the first secondary storage device and a storage area within the second secondary storage device to match data stored in each storage area which is a subject of the copying, wherein, responsive to generation update by the generation setting unit, the copy control unit postpones sending of the second copy request corresponding to the host request belonging to a new generation, and sends to the second secondary storage device all the second copy requests corresponding to the host requests belonging to the old generation which is one generation previous to the new generation, and the secondary copy unit has a first secondary copy mode including: specifying a different part where data is different between the first secondary storage area and the second secondary storage area by using only the newest first history information representing the history of the new generation; and copying only differential data to match data respectively stored in the first secondary storage area and the second secondary storage area, the differential data being data of the different part.

3. A storage system recited in claim 2, wherein the second secondary storage device further comprises: a backup storage area; and a backup copy unit configured to copy data stored in the second secondary storage area into the backup storage area to backup the second secondary storage area in the backup storage area, wherein, the backup copy unit executes the backup after completion notification is issued of all the second copy requests corresponding to the old generation, and sends backup completion notification to the an external device, the copy control unit starts sending the second copy requests corresponding to the new generation after the backup completion notification is issued, the secondary copy unit has a second secondary copy mode including: specifying a different part where data is different between the first secondary storage area and the backup storage area by using history information including at least the newest first history information out of the newest first history information and old first history information representing the history of the generation just one older than the newest first history information; and copying only differential data to match data respectively stored in the first secondary storage area and the backup storage area, the differential data being data of the different part.

4. A storage system recited in claim 3, wherein the backup copy unit is able to send backup generation information to the first secondary storage device responsive to a request, the backup generation information being related to the generation corresponding to the second copy request already reflected in the backup storage area, the secondary copy unit, for the second secondary copy mode, receives the backup generation information from the backup copy unit, and specifies the different part by using only history information whose generation is newer than the generation corresponding to the second copy request already reflected in the backup storage area selected from among the two first history information.

5. A storage system recited in claim 3, wherein when specified first mode conditions are established, the secondary copy unit executes the secondary copy process in the first secondary copy mode, the specified first mode conditions including a first timing condition representing that a point of execution of the secondary copy process is during time from after the completion notification is issued for all the second copy requests corresponding to the old generation and until the start of sending the second copy requests belonging to the new generation, and when specified second mode conditions are established, the secondary copy unit executes the secondary copy process in the second secondary copy mode, the specified second mode conditions including a second timing condition representing that a point of execution of the secondary copy process is during time from after the backup completion notification is issued and until the start of the new backup process.

6. A storage system recited in claim 5, wherein the second secondary storage device further comprises a second history creation unit configured to create second history information representing a write history according to the second copy requests, the secondary copy unit has a third secondary copy mode including: selecting the second secondary storage area when the first timing condition is established; selecting the backup storage area when the second timing condition is established; specifying a different part where data is different between the selected storage area and the first secondary storage area by using at least the second history information; and copying only differential data to match data respectively stored in the selected storage area and the first secondary storage area, the differential data being data of the different part, when the first mode conditions and the second mode conditions are not established, the secondary copy unit executes the secondary copy process in the third secondary copy mode.

7. A storage system recited in claim 5, wherein the second secondary storage device has a status storage area configured to store area status information representing a status of the second secondary storage area, the copy control unit executes sending a first status write request for the status storage area to the second secondary storage device responsive to issuing of all the completion notification for the second copy requests corresponding to the old generation, the first status write request including area status information representing that the first timing condition is established, and sending a second status write request for the status storage area to the second secondary storage device after sending the first status write request and before starting sending of the second copy request belonging to the new generation, the second status write request including area status information representing that the first timing condition is not established, the secondary copy unit judges whether either the first timing condition and the second timing condition is established by referencing the area status information when executing the secondary copy process.

8. A storage system recited in claim 5, wherein the copy control unit executes sending a first notification to the second secondary storage device after issuing of all the completion notification for the second copy requests corresponding to the old generation, the first notification representing that the first timing condition is established, sending a second notification to the second secondary storage device before starting sending of the second copy requests belonging to the new generation, the second notification representing that the first timing condition is not established, the secondary copy unit, when executing the secondary copy process, makes a request to the second secondary storage device for timing condition information which is information relating to which of the first timing condition and the second timing condition is established, the backup copy unit sends to the primary storage device the timing condition information based on the notification received last among the first notification and the second notification and on an execution status of the backup process.

9. A storage system recited in claim 2, wherein the primary storage device has a plurality of primary storage areas identified by mutually different identifiers, the plurality of primary storage areas constitutes a consistency group which are managed with a common generation, the generation setting unit simultaneously updates the generation relating to the plurality of primary storage areas contained in the consistency group.

10. A storage system recited in claim 2, further comprising: a plurality of the primary storage devices; and a management device connected to the host computer and to each of the plurality of the primary storage devices, wherein, the plurality of the primary storage devices have a plurality of the primary storage areas, respectively, which are identified by mutually different identifiers, the plurality of the primary storage areas constitute a consistency group which are managed with a common generation, the management device executes sending to the host computer an instruction for postponing sending of a new host request, responsive to receiving an acknowledgement of the postponing instruction from the host computer, sending instructions to update the generation relating to each of the primary storage areas included in the consistency group to the plurality of the primary storage devices, respectively, and responsive to receiving completion notifications of the generation update instruction from the plurality of primary storage devices respectively, sending to the host computer an instruction to restart sending of the new host request.

11. A storage system recited in claim 2, further comprising: a plurality of the primary storage devices; and a management device connected to the host computer and to each of the plurality of the primary storage devices, wherein, the plurality of the primary storage devices have a plurality of the primary storage areas, respectively, which are identified by mutually different identifiers, the plurality of the primary storage areas constitute a consistency group which are managed with a common generation, the management device executes sending respectively to the plurality of the primary storage devices instructions for postponing execution of the primary write process and the first and second copy request sending process according to a new host request relating respectively to each of the primary storage areas included in the consistency group, responsive to receiving acknowledgements of the postponing instructions from the plurality of primary storage devices respectively, sending instructions to update the generation relating to each of the primary storage areas included in the consistency group to the plurality of the primary storage devices respectively, and responsive to receiving completion notifications of the update instructions from the plurality of primary storage devices respectively, sending respectively to the plurality of the primary storage devices instructions for restarting execution of each of the processes according to the new host request relating respectively to each of the primary storage areas included in the consistency group.

12. A method of controlling a storage system comprising a primary storage device connected to a host computer and having a primary storage area for the host computer, a first secondary storage device connected to the primary storage device and having a first secondary storage area, and a second secondary storage device connected to the primary storage device and having a second secondary storage area, the method comprising the steps of: (A) setting a correlation between the second secondary storage area and a virtual identifier used to identify the second secondary storage area as a virtual storage area within the primary storage device using the primary storage device; (B) executing a primary write process using the primary storage device, the primary write process including a process of receiving from the host computer a host request which is a data write request of host data, and a process of writing the host data to the primary storage area; (C) sending a first and second copy request to the first and second secondary storage devices, respectively, using the primary storage device to store copy data in the first and second secondary storage areas, the copy data being a copy of the host data, the first and second copy request containing the copy data; and (D) sending a completion notification of the host request to the host computer responsive to receiving a completion notification of the first copy request from the first secondary storage device using the primary storage device, regardless of whether or not receiving a completion notification of the second copy request from the second secondary storage device, wherein the step (C) includes identifying the second secondary storage area using the virtual identifier in the second copy request, using the primary storage device; wherein the step (C) includes the steps of: (C1-1) creating the first copy request using the primary storage device; (C1-2) creating the second copy request identifying the second secondary storage area by using the virtual identifier using the primary storage device; and (C2) sending the first and second copy requests to the first and second secondary storage devices, respectively, using the primary storage device, wherein the step (C2) includes the steps of: replacing the virtual identifier within the second copy request with an actual identifier of the second secondary storage area by referencing the correlation set by the step (A) using the primary storage device; and wherein said sending the second copy request to the second secondary storage device using the primary storage device is performed after the replacement.

13. A method recited in claim 12, wherein the first secondary storage device and the second secondary storage device are connected to each other, and the method further comprises the steps of: (E) allocating a generation to the host request and updating the generation according to specified conditions using the primary storage device, the generation representing a time range in which the primary write process is executed; (F) sending first generation information to the first secondary storage device using the primary storage device, the first generation information representing the generation of the host request that is a source of the first copy request; (G) creating first history information for each generation of host requests which are sources of the first copy requests by using the first generation information using the first secondary storage device, the first history information representing a write history according to the first copy requests; (H) executing a secondary copy process of copying data between a storage area within the first secondary storage device and a storage area within the second secondary storage device to match data stored in each storage area which is a subject of the copying using the first secondary storage device; (I) responsive to generation update by the step (E): postponing sending of the second copy request corresponding to the host request belonging to a new generation using the primary storage device; and sending to the second secondary storage device all the second copy requests corresponding to the host requests belonging to the old generation which is one generation previous to the new generation using the primary storage device, wherein the step (H) includes executing a secondary copy process in a first secondary copy mode including: (H1-1) specifying a different part where data is different between the first secondary storage area and the second secondary storage area by using only the newest first history information representing the history of the new generation using the first secondary storage device; and (H1-2) copying only differential data to match data respectively stored in the first secondary storage area and the second secondary storage area using the first secondary storage device, the differential data being data of the different part.

14. A method recited in claim 13, wherein the second secondary storage device further comprises a backup storage area, and the method further comprises the steps of: (J) executing a backup process of copying data stored in the second secondary storage area into the backup storage area to backup the second secondary storage area in the backup storage area after completion notification is issued of all the second copy requests corresponding to the old generation, and sending backup completion notification to the an external device using the second secondary storage device; and (K) starting sending the second copy requests corresponding to the new generation after the backup completion notification is issued using the primary storage device, wherein the step (H) includes executing a secondary copy process in a second secondary copy mode including: (H2-1) specifying a different part where data is different between the first secondary storage area and the backup storage area by using history information including at least the newest first history information out of the newest first history information and old first history information representing the history of the generation just one older than the newest first history information using the first secondary storage device; and (H2-2) copying only differential data to match data respectively stored in the first secondary storage area and the backup storage area using the first secondary storage device, the differential data being data of the different part.

15. A method recited in claim 14, wherein the step (H2-1) includes the steps of: receiving backup generation information from the second secondary storage device using the first secondary storage device, the backup generation information being related to the generation corresponding to the second copy request already reflected in the backup storage area; and specifying the different part by using only history information whose generation is newer than the generation corresponding to the second copy request already reflected in the backup storage area selected from among the two first history information using the first secondary storage device.

16. A method recited in claim 14, wherein the step (H) includes the steps of: executing the secondary copy process in the first secondary copy mode using the first secondary storage device when specified first mode conditions are established, the specified first mode conditions including a first timing condition representing that a point of execution of the secondary copy process is during time from after the completion notification is issued for all the second copy requests corresponding to the old generation and until the start of sending the second copy requests belonging to the new generation; and executing the secondary copy process in the second secondary copy mode using the first secondary storage device, when specified second mode conditions are established, the specified second mode conditions including a second timing condition representing that a point of execution of the secondary copy process is during time from after the backup completion notification is issued and until the start of the new backup process.

17. A method recited in claim 16, further comprising (L) creating second history information representing a write history according to the second copy requests using the second secondary storage device, wherein the step (H) includes when the first mode conditions and the second mode conditions are not established, executing a secondary copy process in a third secondary copy mode including: (H3-1) selecting the second secondary storage area when the first timing condition is established using the first secondary storage device; (H3-2) selecting the backup storage area when the second timing condition is established using the first secondary storage device; (H3-3) specifying a different part where data is different between the selected storage area and the first secondary storage area by using at least the second history information using the first secondary storage device; and (H3-4) copying only differential data to match data respectively stored in the selected storage area and the first secondary storage area using the first secondary storage device, the differential data being data of the different part.

18. A method recited in claim 16, wherein the second secondary storage device has a status storage area configured to store area status information representing a status of the second secondary storage area, the method further comprises the steps of: (M) responsive to issuing of all the completion notification for the second copy requests corresponding to the old generation, sending a first status write request for the status storage area to the second secondary storage device using the primary storage device, the first status write request including area status information representing that the first timing condition is established; (N) sending a second status write request for the status storage area to the second secondary storage device using the primary storage device after sending the first status write request and before starting sending of the second copy request belonging to the new generation, the second status write request including area status information representing that the first timing condition is not established; and (O) judging whether either the first timing condition and the second timing condition is established by referencing the area status information using the first secondary storage device when executing the secondary copy process.

19. A method recited in claim 16, further comprising the steps of: (M) sending a first notification to the second secondary storage device using the primary storage device after issuing of all the completion notification for the second copy requests corresponding to the old generation, the first notification representing that the first timing condition is established; (N) sending a second notification to the second secondary storage device using the primary storage device before starting sending of the second copy requests belonging to the new generation, the second notification representing that the first timing condition is not established; (O) making a request to the second secondary storage device for timing condition information using the first secondary storage device when executing the secondary copy process, the timing condition information relating to which of the first timing condition and the second timing condition is established; and (P) sending to the primary storage device the timing condition information based on the notification received last among the first notification and the second notification and on an execution status of the backup process using the second secondary storage device.

20. A method recited in claim 13, wherein the primary storage device has a plurality of primary storage areas identified by mutually different identifiers, the plurality of primary storage areas constitutes a consistency group which are managed with a common generation, the method further comprises the step of (Q) updating, simultaneously, the generation relating to the plurality of primary storage areas contained in the consistency group using the primary storage device.

21. A method recited in claim 13, wherein the storage system further comprises: a plurality of the primary storage devices; and a management device connected to the host computer and to each of the plurality of the primary storage devices, wherein, the plurality of the primary storage devices have a plurality of the primary storage areas, respectively, which are identified by mutually different identifiers, the plurality of the primary storage areas constitute a consistency group which are managed with a common generation, the method further comprises the steps of (Q) sending to the host computer an instruction for postponing sending of a new host request using the management device; (R) responsive to receiving an acknowledgement of the postponing instruction from the host computer, sending instructions to update the generation relating to each of the primary storage areas included in the consistency group to the plurality of the primary storage devices, respectively using the management device; and (S) responsive to receiving completion notifications of the generation update instruction from the plurality of primary storage devices respectively, sending to the host computer an instruction to restart sending of the new host request using the management device.

22. A method recited in claim 13, wherein the storage system further comprises: a plurality of the primary storage devices; and a management device connected to the host computer and to each of the plurality of the primary storage devices, wherein, the plurality of the primary storage devices have a plurality of the primary storage areas, respectively, which are identified by mutually different identifiers, the plurality of the primary storage areas constitute a consistency group which are managed with a common generation, the method further comprises the steps of: (Q) sending respectively to the plurality of the primary storage devices instructions for postponing execution of the primary write process and the first and second copy request sending process according to a new host request relating respectively to each of the primary storage areas included in the consistency group using the management device; (R) responsive to receiving acknowledgements of the postponing instructions from the plurality of primary storage devices respectively, sending instructions to update the generation relating to each of the primary storage areas included in the consistency group to the plurality of the primary storage devices respectively using the management device; and (S) responsive to receiving completion notifications of the update instructions from the plurality of primary storage devices respectively, sending respectively to the plurality of the primary storage devices instructions for restarting execution of each of the processes according to the new host request relating respectively to each of the primary storage areas included in the consistency group using the management device.

CROSS-REFERENCE TO RELATED APPLICATIONS

The present application claims the priority based on Japanese Patent Application No. 2005-253352 filed on Sep. 1, 2005, the disclosure of which is hereby incorporated herein by reference in its entirety.

BACKGROUND

1. Field of the Invention

The present invention relates to a storage system for copying data and storing in a plurality of storage devices.

2. Description of the Related Art

Remote copy technology is known for a data processing system comprising a storage system having a plurality of storage devices, and a host computer (See, for example, JP2005-78453A, JP2004-5370A, and JP2000-242437A). Remote copy is technology for copying data stored in a storage area included in one storage device to a storage area included in another storage device within the storage system. By using this remote copy technology, it is possible to continue work on the data processing system by using the data stored in the other storage device, even when a problem occurs with the one storage device.

Meanwhile, setting of the copy destination storage area is performed by specifying the storage device having the storage area and that storage area respectively. However, the settings in many cases require a great deal of effort by the user. Also, not enough study had been done for improving the convenience when using a plurality of storage devices for storing copy data.

SUMMARY

An object of the present invention is to provide technology capable of reducing the effort of users required for setting the copy destination storage area. Another object is to provide a technology capable of improving convenience when using a plurality of storage devices for storing copy data.

In an aspect of the present invention, there is provided a storage system for providing a host computer with a data storage area. The storage system comprises a primary storage device, a first secondary storage device, and a second secondary storage device. The primary storage device is connected to the host computer and has a primary storage area for the host computer. The first secondary storage device is connected to the primary storage device and has a first secondary storage area. The second secondary storage device is connected to the primary storage device and has a second secondary storage area. Furthermore, the primary storage device comprises a virtualization unit, a primary writing unit, and a copy control unit. The virtualization unit is configured to set a correlation between the second secondary storage area and a virtual identifier used to identify the second secondary storage area as a virtual storage area within the primary storage device. The primary writing unit is configured to execute a primary write process, the primary write process including a process of receiving from the host computer a host request which is a data write request of host data, and a process of writing the host data to the primary storage area. The copy control unit is configured to send a first and second copy request to the first and second secondary storage devices, respectively, to store copy data in the first and second secondary storage areas, the copy data being a copy of the host data, the first and second copy request containing the copy data. Furthermore, the copy control unit, regardless of whether or not receiving a completion notification of the second copy request from the second secondary storage device, sends a completion notification of the host request to the host computer responsive to receiving a completion notification of the first copy request from the first secondary storage device. The copy control unit identifies the second secondary storage area using the virtual identifier in the second copy request.

With this storage system, copy data is stored in the first secondary storage device, and the primary storage device sends to the host computer the completion notification of the host request responsive to receiving the first copy request completion notification from the first secondary storage device, so it is possible to increase the data redundancy. Also, copy data is stored in the second secondary storage device, and the primary storage device sends to the host computer the host request completion notification, regardless of whether or not the second copy request completion notification has been received from the second secondary storage device, so regardless of the distance from the primary storage device to the second secondary storage device, it is possible to suppress the excess lengthening of the response time for the host computer. Furthermore, the copy control unit identifies the second secondary storage area using the virtual identifier used to identify the second secondary storage area as the virtual storage area within the primary storage device. Therefore, once the virtual identifier is allocated to the second secondary storage area, by using the virtual identifier after that, it is possible to identify the second secondary storage area as the internal area. So it is possible to decrease the effort of the user required for setting the copy destination storage area.

The copy control unit comprises a copy instruction unit and an access control unit. The copy instruction unit is configured to create the first and second copy request, wherein the copy instruction unit creates the second copy request identifying the second secondary storage area using the virtual identifier. The access control unit is configured to send the first and second copy requests to the first and second secondary storage devices, respectively. Furthermore, the access control unit, by referencing the correlation set by the virtualization unit, replaces the virtual identifier within the second copy request with an actual identifier of the second secondary storage area, and sends the second copy request after the replacement to the second secondary storage device.

In this case, the first secondary storage device and the second secondary storage device may be connected to each other. Furthermore the primary storage device may comprise a generation setting unit, and a generation information sending unit. The generation setting unit is configured to allocate a generation to the host request and to update the generation according to specified conditions, the generation representing a time range in which the primary write process is executed. The generation information sending unit is configured to send first generation information to the first secondary storage device, the first generation information representing the generation of the host request that is a source of the first copy request Furthermore the first secondary storage device may comprise a first history creation unit and a secondary copy unit. The first history creation unit is configured to create first history information for each generation of host requests which are sources of the first copy requests by using the first generation information, the first history information representing a write history according to the first copy requests. The secondary copy unit is configured to execute a secondary copy process of copying data between a storage area within the first secondary storage device and a storage area within the second secondary storage device to match data stored in each storage area which is a subject of the copying. Then responsive to generation update by the generation setting unit, the copy control unit postpones sending of the second copy request corresponding to the host request belonging to a new generation, and sends to the second secondary storage device all the second copy requests corresponding to the host requests belonging to the old generation which is one generation previous to the new generation. Furthermore, the secondary copy unit has a first secondary copy mode. The first secondary copy mode includes: specifying a different part where data is different between the first secondary storage area and the second secondary storage area by using only the newest first history information representing the history of the new generation; and copying only differential data to match data respectively stored in the first secondary storage area and the second secondary storage area, the differential data being data of the different part.

With this constitution, in the first secondary copy mode, by copying only the differential data specified by using only the newest first history information, the data stored in the respective storage areas of the first secondary storage device and the second secondary storage device match. So it is possible to improve the convenience when using a plurality of storage devices for storing copy data.

This invention may be embodied in various ways, for example, a storage area providing method and storage area providing device, a storage system control method and storage system control device, a data processing system control method and data processing system control device, a computer program for realizing the functions of such a method or device, a storage medium having such a computer program stored thereon, a data signal embedded in a carrier wave containing such a computer program, or the like.

These and other objects, features, aspects, and, advantages of the present invention will become more apparent from the following detailed description of the preferred embodiments with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is an explanatory drawing showing the hardware constitution of the data processing system as one embodiment of the present invention;

FIG. 2 is a schematic diagram showing the constitution of the data processing system 10 ;

FIG. 3 is an explanatory drawing showing an example of the pair information 248 P;

FIG. 4 is an explanatory drawing showing an example of the volume information 246 P;

FIG. 5 is a block diagram showing the overview of the data writing process by the data processing system 10 ;

FIG. 6 is a schematic diagram showing the data processing system 10 when a problem occurs with the primary device 200 P;

FIG. 7 is an explanatory drawing showing the constitution of the data processing system 10 a for the second embodiment;

FIG. 8 shows the process flow when the primary host 100 P sends the three host requests WR 1 , WR 2 , and WR 3 ;

FIG. 9 is a sequence drawing showing the procedure of the data writing process by the data processing system 10 a;

FIG. 10 is a sequence drawing showing the procedure of the data writing process by the data processing system 10 a;

FIG. 11 is a sequence drawing showing the procedure of the data writing process by the data processing system 10 a;

FIG. 12 is an explanatory drawing showing the copy requests sent to the synchronous secondary device 200 L by the primary device 200 P;

FIG. 13 is an explanatory drawing showing the bitmaps BMA and BMB;

FIGS. 14(A)-14(F) are explanatory drawings showing the sequence in which the synchronous secondary device 200 L receives the requests WR 31 L to WR 35 L and the marker MK;

FIG. 15 is an explanatory drawing showing the status 1 of the data processing system 10 a;

FIG. 16 is an explanatory drawing showing the status 2 ;

FIG. 17 is an explanatory drawing showing the status 3 ;

FIG. 18 is an explanatory drawing showing the status 4 ;

FIG. 19 is an explanatory drawing showing the status 5 ;

FIG. 20 is an explanatory drawing showing the status 6 ;

FIG. 21 is an explanatory drawing showing the status 7 ;

FIG. 22 is an explanatory drawing showing the status 8 ;

FIG. 23 is an explanatory drawing showing the data processing system 10 a when a problem occurs with the primary device 200 P;

FIG. 24 is a flow chart showing the procedure of the secondary copy process;

FIGS. 25(A) and 25(B) are explanatory drawings showing the conditions used for the secondary copy process;

FIG. 26 is an explanatory drawing showing the process executed by the copy module 232 P of the third embodiment when starting the resynchronization of the asynchronous copy pair CP 2 ;

FIG. 27 is a sequence drawing showing the procedure of the data write process by the data processing system 10 a;

FIG. 28 is a sequence drawing showing the procedure of the data write process by the data processing system 10 a;

FIG. 29 is an explanatory drawing showing the status 21 ;

FIG. 30 is an explanatory drawing showing the status 22 ;

FIG. 31 is an explanatory drawing showing the status 23 ;

FIG. 32 is an explanatory drawing showing the status 24 ;

FIG. 33 is an explanatory drawing showing the status 25 ;

FIG. 34 is an explanatory drawing showing the conditions for selecting the differential bitmap;

FIG. 35 is a flow chart showing the procedure of the process for handling a problem with the synchronous copy executed by the primary device 200 P for the fourth embodiment;

FIG. 36 is a schematic diagram showing the constitution of the data processing system 10 b for the fifth embodiment;

FIG. 37 is a sequence drawing showing the procedure of the write process of the status information 280 R;

FIG. 38 is a sequence drawing showing the procedure of the write process of the status information 280 R;

FIG. 39 is a sequence drawing showing the procedure of the write process of the status information 280 R;

FIGS. 40(A) and 40(B) are explanatory drawings showing the conditions for selecting the consistency volume and the conditions for selecting the differential bitmap;

FIG. 41 is an explanatory drawing showing the constitution of the data processing system 10 c for the sixth embodiment;

FIG. 42 is a sequence drawing showing the comparative example of the data write process;

FIG. 43 is a sequence drawing showing the data write process for the sixth embodiment;

FIG. 44 is an explanatory drawing showing the constitution of the data processing system 10 d for the seventh embodiment;

FIG. 45 is a sequence drawing showing the comparative example of the data write process;

FIG. 46 is a sequence drawing showing the data write process for the seventh embodiment;

FIG. 47 is an explanatory drawing showing the constitution of the data processing system 10 e for the eighth embodiment; and

FIG. 48 is a sequence drawing showing the data write process for the eighth embodiment.

DESCRIPTION OF THE PREFERRED EMBODIMENT

Next, the present invention is described based on embodiments in the following sequence.

• A. First embodiment:
• B. Second embodiment:
• C. Third embodiment:
• D. Fourth embodiment:
• E. Fifth embodiment:
• F. Sixth embodiment:
• G. Seventh embodiment:
• H. Eighth embodiment:
• I. Variation examples:

A. First Embodiment

A1. Device Constitution

FIG. 1 is an explanatory drawing showing the hardware constitution of the data processing system as an embodiment of the present invention. This data processing system 10 comprises two host computers 100 ( 100 P, 100 L), three storage devices 200 ( 200 P, 200 L, and 200 R), and a management server 110 P. Note that the overall three storage devices 200 ( 200 P, 200 L, and 200 R) and the management server 110 P correspond to the “storage system.”

The primary host computer 100 P, the primary storage device 200 P, and the management server 110 P are installed at the production site for performing the data processing operation. The secondary host computer 100 L and the secondary storage device 200 L are installed at the local site near the production site. The secondary storage device 200 R is installed at a remote site that is distant from the production site. Hereafter, the production site is also called the “primary site.”

With this specification, a code identifying the site at which an item exists is added to the end of the code representing the host computer and storage device, each structural element, and each type of information and program. Specifically, at the end of the codes representing each of these, the code “P” is added to items existing at the production site, the code “L” is added to items at the local site, and the code “R” is added to items at the remote site. Also, with the description in this specification, when it is not necessary to distinguish the individual host computers, the individual storage devices, or the like, and the code for identifying sites is omitted.

The storage device 200 comprises the control device 210 and the disk array 270 connected to the control device 210 . The control device 210 controls data transfer between the disk array 270 and other devices (for example, the host computer 100 and the other storage device 200 ). The control device 210 comprises the CPU 220 , the memory 230 , and the cache memory 260 . The cache memory 260 is semiconductor memory for temporarily storing data transferred between the control device 210 and the disk array 270 , and is separate from the CPU 220 exclusive cache memory (not illustrated).

The disk array 270 is a device using a plurality of disk devices, and comprises at least one volume 272 . The volume 272 is a logical storage area for storing data used for data processing of the host computer 100 and the like. For example, one volume 272 is formed by logically dividing one logical storage area formed by a plurality of disk devices.

The primary host computer 100 P of the primary site is connected to the primary storage device 200 P (control device 210 P) of the primary site. At normal times, this primary host computer 100 P executes a specified data process while using storage areas provided by the primary storage device 200 P. Data processes include, for example, a process as a file server for providing data files to the client device (not illustrated), or a process as a database server for managing various kinds of data.

The local site secondary host computer 100 L is connected to the local site secondary storage device 200 L (control device 210 L). This secondary host computer 100 L is able to execute specified data processes instead of the primary host computer 100 P. For example, the secondary host computer 100 L normally doesn't execute anything, but when a problem occurs with the primary host computer 100 P, it executes specified data processes.

The storage devices 200 of each site (control device 210 ) are mutually connected via the storage network SN. Used as the storage network SN is a fiber channel or an IP network, for example. The data stored in the primary storage device 200 P by the primary host computer 100 P is copied into each secondary storage device 200 L and 200 R via this storage network SN. By using this kind of copying, it is possible for three geographically separated storage devices 200 to store the same data.

Also, the storage devices 200 of each site (control device 210 ) are connected to the management server 110 P via the management network MN 1 . The management server 110 P controls the operation of each storage device 200 via this management network MN 1 . Also, the management server 110 P comprises the CPU 112 P and the memory 114 P. The user of the data processing system 10 can establish setting of the operation of each storage device 200 by operating this management server 110 P.

The management server 110 P and the primary host computer 100 P are mutually connected via another management network MN 2 . The primary host computer 100 P is able to fetch information relating to the storage area (an identifier for identifying the volume 272 , for example) from the management server 110 P. Note that as each of the management networks MN 1 and MN 2 , an IP network or the like is used.

FIG. 2 is a schematic drawing showing the constitution of the data processing system 10 . The memory 230 of the storage device 200 of each site stores the read/write module 235 . The read/write module 235 executes the write process and the read process for the volume (also called the “internal volume” hereafter) within its own storage device according to a request from outside. With the example in FIG. 2, the primary storage device 200 P comprises an internal volume VOL 1 -P, the secondary storage device 200 L comprises an internal volume VOL 1 -L, and the secondary storage device 200 R comprises the internal volume VOL 1 -R. The volume VOL 3 -P of the primary storage device 200 P is described later.

The memories 230 P and 230 L of the storage devices 200 P and 200 L of the primary site and the local site further store the virtualization module 231 and the copy module 232 . The virtualization module 231 allocates to another storage device volume 272 (FIG. 1) a virtual identifier used for specifying as its own virtual internal volume. The copy module 232 executes data copy processing between the two volumes. Note that the copy module 232 comprises the instruction module 233 and the access module 234 (with the copy module 232 L, the illustration is omitted). Furthermore, the memory 230 P of the primary storage device 200 P of the primary site stores the volume information 246 P and the pair information 248 P. The details of these are described later.

The memory 114 P of the management server 110 P stores the management module 116 P. The management module 116 P executes the process of controlling each storage device 200 .

Note that the functions of each module are realized by the CPU 112 and 220 (FIG. 1) executing the programs (modules).

FIG. 3 is an explanatory drawing showing an example of the pair information 248 P. The pair information 248 P includes the information for defining a copy pair. A copy pair means a combination of two volumes on which data copying is performed. Data stored in one volume constituting the copy pair (called “the copy source volume” or simply the “source volume”) is copied to the other volume (called the “copy destination volume” or simply the “destination volume”).

As shown in FIG. 3, the source volume and the destination volume are specified by an identifier for identifying the storage device 200 (device identifier) and an identifier for identifying the volume (volume identifier). The device identifier is specific to each storage device 200 (more precisely, the interface (not illustrated) for connecting to the storage network SN). The volume identifier is an identifier that can be set independently for each storage device 200 . Furthermore the volume identifier, within one storage device 200 , is specific to each volume. With the storage network SN (FIG. 1), one volume is identified by this kind of combination of the device identifier and the volume identifier.

In FIG. 3, the storage device identifier “P” represents the primary storage device 200 P, the storage device identifier “L” represents the secondary storage device 200 L, and the storage device identifier “R” represents the secondary storage device 200 R. Also, with the description below, for example, the volume for which the volume identifier is “VOL 1 ” comprised by the primary storage device 200 P for which the storage device identifier is “P” is called “volume VOL 1 -P.” The same is also true for other volumes.

In FIG. 3, with the first copy pair CP 1 , the source volume is the volume VOL 1 -P of the primary storage device 200 P (FIG. 2), and the destination volume is volume VOL 1 -L of the secondary storage device 200 L. With the second copy pair CP 2 , the storage device identifier is omitted. This means that both volumes are internal volumes of the primary storage device 200 P. Specifically, with the second copy pair CP 2 , the source volume is the volume VOL 1 -P, and the destination volume is the volume VOL 3 -P. Note that as described later, the volume VOL 3 -P is a virtual internal volume for which the actual entity is the volume VOL 1 -R of the secondary storage device 200 R.

The pair information 248 P (FIG. 3) includes the copy type and the pair status of each copy pair. The copy type is set as one of “synchronous” and “asynchronous.” Details regarding synchronous copying and asynchronous copying are described later. The pair status indicates the status of the copy pair. The pair status is set to one of “pair,” “split,” or “resynchronize.” “Pair” represents the status that the update to the source volume is also reflected in the destination volume. “Split” represents the status that the update to the source volume is not reflected in the destination volume. The split status is used in cases of backing up data of the destination volume to another volume, for example. “Resynchronize” represents the status of shifting from the split status to the pair status. With this resynchronize status, the update to the source volume performed during splitting is reflected in the destination volume. After this reflection is completed, the pair status becomes “pair.”

The pair information 248 P is set by the instruction module 233 P. The instruction module 233 P sets the pair information 248 P according to the instructions from the management server 110 P, for example. It is possible to have the instruction module 233 P allow the user to determine each setting of the pair information 248 P. For example, it is possible to have the user operate the management server 110 P or the operating panel (not illustrated) of the primary storage device 200 P, and for the instruction module 233 P to receive the user instructions from the management server 110 P or the operating panel (not illustrated). This is also the same for information used by other modules described later, and is the same for other embodiments described later.

FIG. 4 is an explanatory drawing showing an example of the volume information 246 P. This volume information 246 P stores the correlation of the internal volume identifier, the external device identifier, and the external volume identifier. The external device identifier and the external volume identifier are identifiers for specifying the volumes of another storage device 200 (also called an “external volume”). The internal volume identifier is a virtual identifier for specifying this external volume as a virtual internal volume. The external volume is handled as an internal volume specified by the internal volume identifier. With the example in FIG. 4, the volume VOL 1 -R of the secondary storage device 200 R is handled as the virtual internal volume VOL 3 -P (details described later). This volume information 246 P is set by the virtualization module 231 P. The virtualization module 231 P sets the volume information 246 P according to instructions from the management server 110 P, for example. It is also possible to have the virtualization module 231 P allow the user to set each setting value of the volume information 246 P.

A2. Data Write Process

FIG. 5 is a block diagram showing the summary of the data write process by the data processing system 10 . Note that with the description below, the primary host computer 100 P is also called simply the “primary host 100 P.” The primary storage device 200 P is also called simply the “primary device 200 P.” The secondary storage device 200 L is called the “synchronous secondary storage device 200 L” or simply the “synchronous secondary device 200 L.” The secondary storage device 200 R is also called the “asynchronous secondary storage device 200 R” or simply the “asynchronous secondary device 200 R.”

First, the primary host 100 P sends to the primary device 200 P the write request WR 10 whose subject is the volume VOL 1 -P. The read/write module 235 P that receives this request WR 10 writes the request data to the volume VOL 1 -P and supplies a completion notification C 10 a to the instruction module 233 P. Following, the write request from the primary host 100 P is also called the “host request.” The data for which write is requested by the host request is also called the “host data.” Note that the instruction module 233 P may also be activated during writing of data based on the request WR 10 by the read-write module 235 P. In this case, the read-write module notifies the instruction module 233 P of the activation of the instruction module 233 P instead of the completion notification C 10 a when the request WR 10 is received. By doing this, in this case, execution of remote copying is performed in parallel with data storage by the primary storage device 200 P.

The instruction module 233 P references the host request WR 10 received by the read-write module 235 P, and detects the copy pair whose source volume is the same as the subject volume VOL 1 -P of this host request WR 10 from the pair information 248 P. With the example in FIG. 3, two copy pairs CP 1 and CP 2 are detected. For the purposes of the following description it is assumed that these pairs CP 1 and CP 2 are detected.

Next, the instruction module 233 P creates a copy (write) request WR 10 b according to the detected synchronous copy pair CP 1 , and supplies the created request WR 10 b to the access module 234 P. This request WR 10 b comprises the same host data as the host request WR 10 and the identifier of the VOL 1 -L (identifier of the device and the volume) which is the destination volume of the synchronous copy pair CP 1 . The access module 234 P sends a request WR 10 c including the same information as the request WR 10 b to the synchronous secondary device 200 L identified by the device identifier “L” specified by the request WR 10 b . The read-write module 235 L (FIG. 2) of the synchronous secondary device 200 L sends the completion notification C 10 c of the request WR 10 c to the primary device 200 P when the received host data is written to the specified volume VOL 1 -L. The access module 234 P supplies the completion notification C 10 b of the request WR 10 b to the instruction module 233 P responsive to receiving of the completion notification C 10 c.

Furthermore, the instruction module 233 P creates the copy (write) request WR 10 d according to the detected asynchronous copy pair CP 2 , and supplies the created request WR 10 d to the access module 234 P. This request WR 10 d comprises the same host data as the host request WR 10 and the identifier of the VOL 3 -P (virtual internal volume identifier) which is the destination volume of the asynchronous copy pair CP 2 . By referencing the volume information 246 P (FIG. 4), the access module 234 P converts the virtual identifier of the volume VOL 3 -P within the request WR 10 d to the volume VOL 1 -R identifier (identifier of the device and volume). Then, the access module 234 P sends the request WR 10 e after conversion to the asynchronous secondary device 200 R identified by the device identifier “R” after conversion. The read-write module 235 R (FIG. 2) of the asynchronous secondary device 200 R sends the completion notification C 10 e of the request WR 10 e to the primary device 200 P when the received host data is written to the specified volume VOL 1 -R. The access module 234 P supplies the completion notification C 10 d of the request WR 10 d to the instruction module 233 P responsive to receiving of the completion notification C 10 e.

Also, the instruction module 233 P sends the completion notification C 10 of the host request WR 10 to the primary host 100 P responsive to receiving of the completion notification C 10 a of writing to the source volume VOL 1 -P and receiving of the completion notification C 10 c (C 10 b ) of the request WR 10 c of the synchronous copy pair CP 1 . Also, when sending this completion notification C 10 , there is no consideration of whether or not the instruction module 233 P receives the completion notification C 10 e (C 10 d ) of the request WR 10 e of the asynchronous copy pair CP 2 .

In this way, with the synchronous copying, at the point that the primary host 100 P receives the completion notification C 10 , the host data is stored in both the primary device 200 P (source volume VOL 1 -P) and the synchronous secondary device 200 L (destination volume VOL 1 -L). Therefore, by using synchronous copying, it is possible to increase the redundancy of data. However, to prevent excessive lengthening of the response time for the primary host 100 P, it is preferable to install the destination volume storage device 200 L in an area close to the source volume storage device 200 P (within an area of a distance of up to about 100 km, for example).

Meanwhile, with asynchronous copying, at the point that the primary host 100 P receives the completion notification C 10 , there is no guarantee that the host data is stored in the asynchronous secondary device 200 R (destination volume VOL 3 -P (VOL 1 -R)). However, the completion notification is sent to the primary host 100 P without waiting for the copy request completion notification, so it is possible to shorten the response time for the primary host 100 P. Therefore, it is possible to install the destination volume storage device 200 R without restricting the distance from the source volume storage device 200 P.

Note that the instruction module 233 P is able to execute sending of the asynchronous copy request at any time. For example, the instruction module 233 P is able to regularly send asynchronous copy requests. In this case, the instruction module 233 P stores an update history (not illustrated) of the source volume VOL 1 -P in a memory (the memory 230 P or the cache memory 260 P, for example), and requests are supplied to the access module 234 P according to this update history.

FIG. 6 is a schematic diagram showing the data processing system 10 when a problem occurs with the primary device 200 P. In this case, a copy of the data stored in the volume VOL 1 -P is stored in the volume VOL 1 -L. The secondary host 100 L can restart data processing quickly by using this volume VOL 1 -L.

Also, the synchronous secondary device 200 L comprises the same virtualization module 231 L and the same copy module 232 L as the primary device 200 . Therefore, the same as with the primary device 200 P, the virtualization module 231 L sets the volume information 246 L for virtualizing the volume VOL 1 -R, and the copy module 232 L can set the pair information 248 L using this virtual internal volume (with the example in FIG. 6, the volume VOL 4 -L). As a result, the synchronous secondary device 200 L, the same as with the primary device 200 P, is able to store in the volume VOL 1 -R a copy of the data stored in the volume VOL 1 -L. However, it is also possible to omit these modules 231 L and 232 L.

Furthermore, when a problem occurs in the synchronous secondary device 200 L as well, the host computer (not illustrated) can be connected to the asynchronous secondary device 200 R. This host computer can restart the data process using the data stored in the volume VOL 1 -R.

As described above, with the data processing system 10 of the first embodiment, a copy of the data stored in the primary device 200 P is also stored in the two secondary storage devices 200 L and 200 R, so it is possible to increase the redundancy of the data. Furthermore, the synchronous copying is executed to the synchronous secondary device 200 L which is relatively close to the primary device 200 P, and the asynchronous copying is executed to the asynchronous secondary device 200 R which is relatively far from the primary device 200 P. Therefore, it is possible to prevent excessive lengthening of the response time for the primary host 100 P.

Furthermore, with the data processing system 10 of the first embodiment, the access module 234 P (FIG. 2, FIG. 5) receives the copy request for the virtual internal volume (the request WR 10 d , for example), and can send the copy request (the request WR 10 e , for example) for the actual entity volume. Therefore, the instruction module 233 P is able to determine the copy pair (FIG. 3) without specifying the asynchronous secondary device 200 R. For example, once the volume information 246 P is set, after that, even if the user does not know the identifier to identify the secondary storage device 200 R, it is possible for the user to set the pair information 248 P. Therefore, it is possible to decrease the effort by the user for constructing a storage system using the asynchronous secondary device 200 R installed at a remote location. Furthermore, it is preferable that the instruction module 233 P present to the user a list of usable internal volumes (including virtual internal volumes) via the operating panel (not illustrated) or the management server 110 P. By doing this, it is possible for the user to easily determine the copy pair.

Note that among the storage devices 200 (read-write module 235 ), some devices issue the completion notification of copy (write) request when the write requested data is stored in the cache memory 260 , and after that, write the data to the volume. When using this kind of storage device 200 , it is possible to use this kind of completion notification as the completion notifications C 10 a , C 10 c , and C 10 e (FIG. 5). In this way, the completion notification of the copy request (write request) is not restricted to a notification that the actual writing is completed to the volume, but rather has a broad meaning including the notification that the writing is completed to temporarily used memory such as the cache memory 260 .

Also, sending of the completion notification C 10 to the primary host 100 P is not limited to the instruction module 233 P, but rather it is also possible for this to be executed by the access module 234 P or by another module (not illustrated) that the copy module 232 P has.

B. Second Embodiment

FIG. 7 is an explanatory drawing showing the constitution of the data processing system 10 a of the second embodiment. The hardware constitution of the data processing system 10 a is the same as that of the data processing system 10 shown in FIG. 1. However, the functional constitution of each storage device 200 is different from that of the data processing system 10 shown in FIG. 2. The difference for each storage device 200 is as follows.

(1) Difference for the Primary Storage Device 200 P:

The memory 230 P further stores the generation setting module 236 P and the generation information sending module 237 P.

(2) Difference for the Synchronous Storage Device 200 L:

The memory 230 L further stores the history creation module 238 L, the secondary copy module 240 L, the first bitmap BMA, and the second bitmap BMB.

(3) Difference for the Asynchronous Secondary Storage Device 200 R:

The memory 230 R further stores the history creation module 238 R, the backup module 242 R, the pair information 248 R, and the third bitmap BMC. Also, the disk array 270 R (FIG. 1) of the secondary storage device 200 R comprises the volume VOL 2 -R in addition to the volume VOL 1 -R.

The history of updating for the volume VOL 1 -L is stored in the two bitmaps BMA and BMB of the synchronous secondary device 200 L. Meanwhile, the history of updating for the volume VOL 1 -R is stored in the third bitmap BMC of the asynchronous secondary device 200 R. These three bitmaps BMA to BMC are used when a problem occurs with the primary storage device 200 P (details are described later).

FIG. 7 shows an example of the pair information 248 R. The backup copy pair BCP is defined with this pair information 248 R. With this backup copy pair BCP, the source volume is the VOL 1 -R and the destination volume is the VOL 2 -R (both volumes are internal volumes). The backup module 242 R copies the data of the source volume VOL 1 -R to the destination volume VOL 2 -R according to this backup copy pair BCP. Following, the destination volume VOL 2 -R is also called the “backup volume VOL 2 -R.” Also, the pair information 248 R includes the asynchronous copy pair CP 2 destination volume and the pair status (regarding the source volume is omitted). Note that the backup module 242 R sets this pair information 248 R the same as the instruction module 233 P sets the pair information 248 P. However, the information relating to the asynchronous copy pair CP 2 is set according to the instructions of the copy module 232 P.

B1. Host Consistency

With the second embodiment, the process of asynchronous copying from the volume VOL 1 -P to the volume VOL 3 -P (VOL 1 -R) and the process of backup copying from the volume VOL 1 -R to the volume VOL 2 -R while the asynchronous copying is in a suspended state are alternately repeated. The reason that asynchronous secondary device 200 R does a backup copy of data of the volume VOL 1 -R to the separate volume VOL 2 -R is to secure at least one volume having consistency with the host 100 . FIG. 8 and FIG. 9 are sequence drawings showing the procedure of the data write process by the data processing system 10 a . These sequence drawings are for describing the “consistency with the host 100 ,” and the backup process is omitted.

FIG. 8 shows the flow of the process when the primary host 100 sends the three host requests WR 1 , WR 2 , and WR 3 . The three host requests WR 1 , WR 2 , and WR 3 are sent in this sequence to the primary device 200 P. With the example in FIG. 8, it is assumed that changing of the data write sequence of these three requests WR 1 , WR 2 , and WR 3 is prohibited. The prohibition of the write sequence occurs, for example, when a plurality of host requests repeatedly update the same data file, or when a plurality of host requests include a host request for “income processing” for a financial database and a host request for “expense processing after that income processing”.

With the description below, the simplified description “the storage device 200 executes processing,” is used, but the process of writing data to volumes, the process of sending copy requests, the process of sending copy request completion notification, and the like, are executed by each module of the storage device 200 , the same as the example shown in FIG. 5. Furthermore, with the description below, a code for identifying the site of the secondary storage device is added at the end of the code representing the copy request (write request) sent to each secondary storage device 200 by the primary device 200 P according to the host request. For example, the request WR 1 L sent to the synchronous secondary device 200 L by the primary device 200 P is a copy request according to the host request WR 1 . Similarly, a code for identifying the site of the secondary storage device 200 is also added to the end of the code representing the completion notification sent to the primary device 200 P by each secondary storage device 200 . This is also the same for other embodiments described later.

With the example in FIG. 8, first, the primary host 100 P sends the host request WR 1 to the primary device 200 P (step H 700 ). When this is done, the same as with the example in FIG. 5, the primary device 200 P and the synchronous secondary device 200 L execute the processing based on synchronous copying. In specific terms, the primary device 200 P executes write processing according to this request WR 1 (P 702 ), and sends the copy request WR 1 L to the synchronous secondary device 200 L (P 704 ). The synchronous secondary device 200 L executes write processing according to this request WR 1 L (L 706 ) and sends the completion notification C 1 L to the primary storage device 200 P (L 708 ). The primary storage device 200 P that receives the completion notification C 1 L sends the completion notification C 1 to the primary host 100 P (P 709 ).

The primary host 100 P sends to the primary device 200 P the next host request WR 2 responsive to receiving the completion notification C 1 of the request WR 1 (step H 710 ). The series of processes SWR 2 (H 710 to P 719 ) according to this request WR 2 is the same as the series of processes SWR 1 (H 700 to P 709 ) according to the request WR 1 . Similarly, the primary host 100 P sends the next host request WR 3 to the primary device 200 P responsive to receiving the completion notification C 2 of the request WR 2 (step H 720 ). The series of processes SWR 3 (H 720 to P 729 ) according to the request WR 3 is also the same as the series of processes SWR 1 and SWR 2 according to each request WR 1 and WR 2 .

As described above, the primary host 100 P sends the next host request to the primary device 200 P responsive to receiving the completion notification of the previous host request from the primary device 200 P. As a result, for both the primary device 200 P and the synchronous secondary device 200 L, there is no changing of the sequence of writing the requests WR 1 , WR 2 , and WR 3 . As a result, even if there is a case when a problem occurs with the primary device 200 P during processing according to these requests WR 1 , WR 2 , and WR 3 , the secondary host 100 L (FIG. 1) is able to restart the data processing by using the synchronous secondary device 200 L. For example, when a problem occurs at the timing T 1 (FIG. 8) between the step L 708 and the step P 709 , the secondary host 100 L is able to restart the data processing from the status at the timing T 1 by using the synchronous secondary device 200 L (volume VOL 1 -L). Hereinafter, the fact that the volume is in the state in which the data processing can be started by the host 100 using that volume as is, is called “that volume has consistency with the host 100 .” Following, “consistency with the host” is also called “host consistency.”

Meanwhile, the primary device 200 P and the asynchronous secondary device 200 R execute processing based on the asynchronous copy. The primary device 200 P sends to the storage device 200 R (step P 730 ) the copy (write) request for the updated part according to the request from the primary host 100 P among the data stored in the volume VOL 1 -P. As a result, the requests WR 1 R, WR 2 R, and WR 3 R are sent to the asynchronous secondary device 200 R. Here, the sequence for receiving of the requests WR 1 R, WR 2 R, and WR 3 R by the asynchronous secondary device 200 R may sometimes be different from the sequence in which the requests were sent by the primary device 200 P. This is because there is a plurality of communication paths between the primary device 200 P and the asynchronous secondary device 200 R, and furthermore, because there are cases when the amount of communication delay is different for each communication path (this kind of data transfer using a plurality of paths is also called “multiplex transfer”). Also, with the second embodiment, the asynchronous secondary device 200 R writes the data of each request to the volume VOL 1 -R in the sequence the requests are received. Therefore, before the asynchronous secondary device 200 R issues completion notifications for all of these requests WR 1 R, WR 2 R, and WR 3 R, it is possible that the volume VOL 1 -R does not have host consistency.

For example, with the example in FIG. 8, the primary device 200 P sends each request WR 1 R, WR 2 R, and WR 3 R in sequence, but the asynchronous secondary device 200 R receives in the sequence WR 2 R, WR 3 R, and WR 1 R (steps R 740 , R 742 , and R 744 ). Let us assume that at the timing T 2 between the step R 742 and the step R 744 , a problem occurs with the communication path between the primary device 200 P and the asynchronous secondary device 200 R. When this is the case, the volume VOL 1 -R is in the state of storing the data of the request WR 2 and the request WR 3 without storing the data of the request WR 1 . Here, the secondary host 100 L cannot go back to the past and send the request WR 1 , so it is not possible to restart data processing by using the asynchronous secondary device 200 R (volume VOL 1 -R) as is. Specifically, the volume VOL 1 -R does not have host consistency. Meanwhile, at the timing T 3 after issuing of the completion notification of all the requests WR 1 , WR 2 , and WR 3 , the volume VOL 1 -R has host consistency.

Note that the copy module 232 P (FIG. 7) performs update status management of the volume VOL 1 -P by dividing the volume VOL 1 -P into a plurality of blocks for asynchronous copying. The copy module 232 P stores the information for identifying updated blocks in a memory (e.g. the memory 230 P or the cache memory 260 P). Then, the copy module 232 P sends to the asynchronous secondary device 200 R the write request whose subject is only the updated block data. When each of the requests WR 1 , WR 2 , and WR 3 have the same blocks as subjects, only one write request using the newest data is sent for common blocks. Also, when a plurality of write requests with the same block as the subject are sent to the asynchronous secondary device 200 R, the copy module 232 P sends the later request after receiving the completion notification of the previous request. Therefore, even when the request arrival sequence is irregular at the asynchronous copy destination volume VOL 1 -R, overwriting of the relatively new update data by relatively old update data is prevented. Note that this kind of update status management is performed by the instruction module 233 .

FIG. 9 is a sequence drawing showing another example of the data writing process. With the example in FIG. 9, the primary host 100 P sends to the primary device 200 P the four host requests WRa 1 , WRa 2 , WRb 1 , and WRb 2 . With the example in FIG. 9, it is assumed that changing of the write sequence of the two requests WRa 1 and WRa 2 of the first request group RGa is not allowed. Similarly, changing of the write sequence of the two requests WRb 1 and WRb 2 of the second request group RGb is not allowed. However, changing of the write sequence between the two groups RGa and RGb is allowed. As a case of allowing changing of the write sequence, for example, there is the case that the plurality of requests updates different data files respectively, or the case that the plurality of requests updates different databases respectively.

First, the primary host 100 P sends the request WRa 1 to the primary device 200 P (step H 800 ). The primary device 200 P executes the write processing according to this request WRa 1 (P 802 ), and sends the request WRa 1 L to the synchronous secondary device 200 L (P 804 ). Here, when changing of the data write sequence is allowed, the primary host 100 P sends the next host request before receiving the completion notification of the previous host request. In light of this, the primary host 100 P sends the next request WRb 1 to the primary device 200 P without waiting to receive the completion notification of the previous request WRa 1 (H 806 ). The primary device 200 P executes the write process according to this request WRb 1 (P 808 ), and sends the request WRb 1 L to the synchronous secondary device 200 L (P 810 ).

Here, the sequence in which the synchronous secondary device 200 L receives these requests WRa 1 L and WRb 1 L may differ from the sequence in which the requests WRa 1 L and WRb 1 L are sent. This is because there are cases when the data transfer between the primary device 200 P and the synchronous secondary device 200 L is performed by multiplex transfer. With the example in FIG. 9, the synchronous secondary device 200 L receives the request WRb 1 L sent later before receiving the request WRa 1 L sent earlier. Also, with the second embodiment, the synchronous secondary device 200 L writes the requested data to the volume VOL 1 -L in the sequence in which the requests are received. In light of this, the synchronous secondary device 200 L executes the write process according to the request WRb 1 L (L 812 ), and sends the completion notification Cb 1 L of the request WRb 1 L to the primary device 200 P (L 814 ). The primary device 200 P which has received the completion notification Cb 1 L sends the completion notification Cb 1 of the request WRb 1 to the primary host 100 P (P 816 ). The primary host 100 P sends the next request WRb 2 to the primary device 200 P responsive to receiving this completion notification Cb 1 (H 818 ). This request WRb 2 is sent without waiting for the completion notification Ca 1 of the request WRa 1 . The primary device 200 P executes write processing according to this request WRb 2 (P 820 ) and sends the copy request WRb 2 L to the synchronous secondary device 200 L (P 832 ). The synchronous secondary device 200 L executes the write processing according to the received request WRb 2 (L 834 ).

Also, the synchronous secondary device 200 L executes write processing according to the request WRa 1 L (L 822 ), and sends the completion notification Ca 1 L of the request WRa 1 L to the primary device 200 P (L 824 ). The primary device 200 P that receives the completion notification Ca 1 L sends the completion notification Ca 1 of the request WRa 1 to the primary host 100 P (P 826 ). The primary host 100 P sends the next request WRa 2 to the primary device 200 P after receiving this completion notification Ca 1 (H 828 ). The primary device 200 P executes write processing according to this request WRa 2 (P 830 ), and sends the copy request WRa 2 L to the synchronous secondary device 200 L (P 836 ). The synchronous secondary device 200 L executes the write processing according to the received request WRa 2 L (L 838 ).

As described above, the primary host 100 P sends the next host request before receiving the completion notification of the previous host request for a plurality of host requests for which changing of the data write sequence is allowed. As a result, with the synchronous secondary device 200 L, there are cases when the write sequence of the requests is changed. However, the primary host 100 P sends the next host request after receiving the completion notification of the previous host request for a plurality of host requests for which changing of the data write sequence is not allowed (for example, a combination of the requests WRa 1 and WRa 2 or a combination of the requests WRb 1 and WRb 2 ). Therefore, the volume VOL 1 -L of the storage device 200 L always has host consistency. For example, let us assume that a problem occurred at the communication path between the primary device 200 P and the synchronous secondary device 200 L at the timing T 10 between the step L 812 and the step L 822 of FIG. 9. In this case, the volume VOL 1 -L is in the state of storing the later request WRb 1 data without storing the previous request WRa 1 data. However, the changing of the write sequence of these requests WRa 1 and WRb 1 is allowed. Therefore, the secondary host 100 L (FIG. 1) is able to restart data processing by using the synchronous secondary device 200 L (volume VOL 1 -L) as is.

Note that the primary device 200 P and the asynchronous secondary device 200 R execute processing based on asynchronous copying (steps P 840 , R 842 to R 850 ). These processes are performed in the same way as the processes described with FIG. 8 (P 730 , R 740 to R 744 ).

As described using FIG. 8 and FIG. 9 above, the synchronous copy destination volume VOL 1 -L always has host consistency. Meanwhile, the asynchronous copy destination volume VOL 1 -R has host consistency at the point that the asynchronous copying is completed, but it is possible not to have host consistency during asynchronous copying. In light of this, with the second embodiment, backup of the volume VOL 1 -R is performed using the volume VOL 2 -R for the asynchronous secondary device 200 R to secure at least one volume having host consistency.

B2. Data Write Processing

FIG. 10 and FIG. 11 are sequence drawings showing the procedure of the data write process by the data processing system 10 a . FIG. 11 shows the latter half after the procedure of FIG. 10. These sequence drawings include the volume VOL 1 -R backup process. Furthermore, with this example, the primary host 100 P sends the five host requests WR 31 to WR 35 to the primary device 200 P. Here, changing of the write sequence of these requests WR 31 to WR 35 is allowed. Also, the same as with the example in FIG. 5, each storage device 200 sends the copy (write) request completion notification, but with the description below, the figure and description of this is omitted.

The primary host 100 P sends the requests WR 31 to WR 35 in this sequence to the primary device 200 P. The primary device 200 P executes the write processing and the process of sending copy requests to the synchronous secondary device 200 L according to the requests in the sequence in which the requests were received (the same as the sent sequence) (steps P 100 to P 106 , P 116 , P 118 ).

FIG. 12 is an explanatory drawing showing the copy requests sent to the synchronous secondary device 200 L by the primary device 200 P. Each of the copy requests WR 31 L to WR 35 L includes a generation number GNo, a sequence number SNo, and a host data HD. The sequence number SNo is a number indicating the sequence in which the source host request is received by the read/write module 235 P (FIG. 7), and are allocated to the requests WR 31 L to WR 35 L by the instruction module 233 P. Note that a marker MK is sent between the two requests WR 33 L and WR 34 L (FIG. 10: step P 114 ). This marker MK is described later.

The generation number GNo is a number representing a time range in which the read/write module 235 P (FIG. 5, FIG. 7) receives the source host request and executes the host data write processing. The generation setting module 236 P allocates the generation number GNo to the host request by using the receiving time as the reference for each receiving of a host request by the read/write module 235 P. Also, the generation setting module 236 P switches this generation number to a new number responsive to each receiving of the split instruction (details described later) from the management server 110 P. The generation information sending module 237 P adds this generation number GNo to each of the requests WR 31 L to WR 35 L prepared by the access module 234 P to send the generation number GNo to the synchronous secondary device 200 L. With the example in FIG. 12, the generation number of the first three requests WR 31 L to WR 33 L is set to 100 . The generation number of the next two requests WR 34 L and WR 35 L is set to one newer 101 . This is because the primary device 200 P received a split instruction between the two requests WR 33 and WR 34 (FIG. 10, step P 110 , details described later). Note that the read/write module 235 P executes the host data write process in the sequence in which the host requests were received. Therefore, the generation setting module 236 P can use the time of the start of the write process or the time that the write process completion notification is issued as a reference time for allocating generation numbers.

FIG. 13 is an explanatory drawing showing the bitmaps BMA and BMB. The history creation module 238 L (FIG. 7) of the synchronous secondary device 200 L manages update history by dividing the volume VOL 1 -L into a plurality of blocks. Each bitmap BMA and BMB contains bits representing the presence or absence of updates for each block. With the example in FIG. 13, the history creation module 238 L first sets all the bits to 0, and after that, sets the bits of the updated blocks to 1. Bit setting of the updated blocks is performed with each execution of the synchronous copy write processing by the read/write module 235 L. Note that the number of blocks updated by one request is value that varies according to the host data.

Furthermore, each bitmap BMA and BMB contains generation information GI and history status information HSI. The generation information GI indicates the bitmap generation number GNo (FIG. 12). The history creation module 238 L creates a bitmap for each generation number GNo. Furthermore, the history creation module 238 L secures only the newest two generation bitmaps for the bitmaps BMA and BMB. With the example in FIG. 13, the generation number of the first bitmap BMA is 100 , and the generation number of the second bitmap BMB is 101 . Note that before the synchronous secondary device 200 L receives the request of the generation number 101 , the update history of the generation number 99 is stored in the second bitmap BMB (details described later).

The history status information HSI is information indicating the presence or absence of the possibility of future updates. The history creation module 238 L switches the history status information HSI from “undetermined” to “determined” responsive to issuing of the completion notification for all the copy requests of the corresponding generation number GNo (details described later).

Above, the bitmaps BMA and BMB of the synchronous secondary device 200 L are described, the constitution of the third bitmap BMC (FIG. 7) of the asynchronous secondary device 200 R also is the same as that of these bitmaps BMA and BMB. The history creation module 238 R updates the third bitmap BMC with each execution of the asynchronous copy write processing by the read/write module 235 R.

Note that as will be described later, for the generation information of the third bitmap BMC, an update is done one at a time according to the instructions of the primary device 200 P copy module 232 P. The initial value of the third bitmap BMC generation may be set in advance by various methods. For example, it is possible to have the generation information sending module 237 P send the initial value of the generation to the asynchronous secondary device 200 R (history creation module 238 R) in advance. It is also possible to have the initial value set according to management server 110 P instructions or user instructions.

Next, using the examples in FIG. 10 and FIG. 11, the changing of the status of the data processing system 10 a (FIG. 7) is described. When the management server 110 P (management module 116 P) sends the split instruction to the primary device 200 P (step M 108 ), the primary device 200 P starts the split processing according to the instruction (P 110 ). The generation number is switched to just one newer number responsive to this split instruction. Furthermore, data copying according to the new generation host request is postponed for the asynchronous copy pair (VOL 1 -P, VOL 3 -P). With the example in FIG. 10, this split instruction is received at the primary device 200 P between the two requests WR 33 and WR 34 . The copy module 232 P holds the new generation host requests (WR 34 , 35 ) received after this split instruction in the cache memory 260 P to send these requests later to the asynchronous secondary device 200 R.

The split instruction is sent to back up the asynchronous copy destination volume VOL 3 -P (VOL 1 -R). The management server 110 P is able to send this instruction at any time. For example, the split instruction can be sent periodically. It is also possible to have the split instruction sent at a timing according to user instructions.

The generation setting module 236 P (FIG. 7) switches (step P 112 ) the generation number ( 100 ) to one newer number ( 101 ) responsive to receiving the split instruction. Next, the generation information sending module 237 P sends the marker MK representing the generation switching to the synchronous secondary device 200 L (P 114 ).

FIG. 12 shows an example of the marker MK. The marker MK has the same constitution as the other requests WR 31 L to WR 35 L. The generation number GNo is set to a new generation number ( 101 ). The part correlating to the host data HD is set to a specified value indicating that this is a marker. The sequence number SNo is set to a number indicating the receiving sequence of the split instructions by the primary device 200 P the same as with the write requests WR 31 L to WR 35 L. The marker MK sequence number SNo is set to one newer number ( 1004 ) than the final request (WR 33 L) before receiving of the split instruction. Also, the first request (WR 34 L) sequence number SNo after receiving of the split instruction is set to one newer number ( 1005 ) than the marker MK. Setting of the sequence number SNo is performed by the instruction module 233 P.

The sequence number SNo given to the marker MK is used for judging whether or not the completion notification has been issued for all the requests of one generation previous to the generation of the marker MK. For example, with the example in FIG. 12, the sequence number SNo of the marker MK is 1004 . Therefore, if the completion notification is issued for all the request up to the one prior 1003 by the synchronous secondary device 200 L, it is possible to judge that the completion notification has been issued for all the requests of the generation one prior. This judgment is performed by the history creation module 238 L (FIG. 7) (described later).

FIG. 14 is an explanatory drawing showing the sequence of receiving each request WR 31 L to WR 35 L and the marker MK by the synchronous secondary device 200 L. With the example in FIG. 10, the synchronous secondary device 200 L receives the data in the sequence of WR 32 L, marker MK, WR 35 L, WR 31 L, WR 33 L, and WR 34 L (L 104 to L 126 ). FIG. 14(A) to FIG. 14(F) indicate the data receiving situation in that sequence. This receiving sequence is different from the sending sequence because data transfer is performed by multiplex transfer. The synchronous secondary device 200 L (FIG. 7, read/write module 235 L) writes each request host data to the volume VOL 1 -L in the sequence in which each request was received (steps L 104 to L 126 ). The history creation module 238 L also updates the bitmaps BMA and BMB in the same sequence.

FIG. 15 is an explanatory drawing showing the status of the data processing system 10 a at the stage at which the request WR 32 write process (L 104 ) is executed by the synchronous secondary device 200 L. Shown in FIG. 15 are the four volumes VOL 1 -P, VOL 1 -L, VOL 1 -R, and VOL 2 -R, and the three bitmaps BMA, BMB, and BMC. The numbers in parentheses attached to each volume indicate the generation number GNo of the request reflected in that volume. For example, the generation number 99 request is reflected in the backup volume VOL 2 -R, but the generation number 100 request is not reflected.

Also, the numbers in parentheses added to each bitmap indicate the generation number of that bitmap. Also, in the drawing, a table is shown indicating the generation number and history status of each bitmap. Note that in the drawing, each of the bitmaps BMA, BMB, and BMC is indicated using a code with the code “BM” omitted. For example, the code “A” indicates the first bitmap BMA. The same is also true for the other drawings described later.

Note that the double line arrow in the figure indicates the secondary copy process executed when a problem occurs with the primary device 200 P (details described later).

With this status 1 , the pair status of the synchronous copy pair CP 1 “VOL 1 -P, VOL 1 -L” is “pair.” The pair status of the asynchronous copy pair CP 2 “VOL 1 -P, VOL 1 -R” is also “pair.” The pair status of the backup copy pair BCP “VOL 1 -R, VOL 2 -R” is “split.” For the second bitmap BMB, the generation number is “ 99 ,” and the history status is “determined.” Meanwhile, for the first bitmap BMA and the third bitmap BMC, the generation number is “ 100 ,” and the history status is “undetermined.”

FIG. 16 is an explanatory drawing showing the status 2 . When the primary device 200 P receives the split instruction (FIG. 10, P 110 ), and the synchronous secondary device 200 L executes the write process of the request WR 35 L of the new generation 101 (FIG. 10, L 120 ), the system 10 a status changes to status 2 . The history creation module 238 L references the generation number of the copy request received by the read/write module 235 L. When the referenced generation number is newer than the generation number of both the two bitmaps BMA and BMB, the history creation module 238 L clears the older bitmap, and stores this new generation history in the cleared bitmap. Also, the history status of the cleared bitmap is set to “undetermined.” With the example in FIG. 16, the update of the new generation number 101 is stored in the second bitmap BMB.

FIG. 17 is an explanatory drawing showing the status 3 . When the synchronous secondary device 200 L executes the write process of the requests WR 31 L and WR 33 L (FIG. 10, L 122 , L 124 ), the system 10 a status changes to status 3 . The history creation module 238 L finds out that the sequence number SNo of the copy request of the generation number 100 is up to 1003 by referencing the marker MK received by the synchronous secondary device 200 L. Also, the synchronous secondary device 200 L issues a completion notification for all the requests up to sequence number SNo 1003 at step L 124 . In light of this, the history creation module 238 L judges that the completion notification was issued for all the copy requests of the generation number 100 at this step L 124 , and sets the history status of the bitmap BMA of the generation number 100 to “determined.”

After that, the copy module 232 P (FIG. 7) of the primary device 200 P sends the copy request based on the asynchronous copy to the asynchronous secondary device 200 R (FIG. 10, P 128 ). The requests sent here are only items based on all the requests WR 31 , WR 32 , and WR 33 of the generation number 100 before the split instruction. The asynchronous secondary device 200 R executes the write processing according to these requests (R 130 ), and sends the completion notification to the primary device 200 P (R 131 ). Note that hereafter, the description is of all the asynchronous remote copy data (here, WR 31 , 32 , and 33 ) being sent to the asynchronous secondary storage device 200 R at the time of the split instruction, but the data transfer timing is not limited to this. For example, it is also possible to execute asynchronous remote copying at any time. Furthermore, when a split instruction is received, the primary storage device 200 P may confirm the presence or absence of requests WR for which generation numbers have been allocated before the split instruction, and send the applicable requests WR to the asynchronous secondary device 200 R.

After issuing of the completion notification for all the copy requests based on the asynchronous copy, the system 10 a status changes to status 4 . FIG. 18 is an explanatory drawing showing the status 4 . The copy module 232 P sets the pair status of the asynchronous copy pair CP 2 for the pair information 248 P to “split” (P 132 ) responsive to receiving the completion notification of all the copy requests based on the asynchronous copy from the asynchronous secondary device 200 R. Then, the copy module 232 P sends the split completion notification to the asynchronous secondary device 200 R. Next, at step R 143 , at the asynchronous secondary device 200 R, the history creation module 238 R that received the split completion notification sets the history status of the third bitmap BMC to “determined.” Also, the backup module 242 R that received the split completion notification sets the pair status of the asynchronous copy pair CP 2 for the pair information 248 R (FIG. 7) to “split.” As a result, the system 10 a status becomes status 4 for which the split is completed. With this status 4 , the asynchronous cop destination volume VOL 1 -R has host consistency.

After completion of the split, the system 10 a status changes to status 5 . FIG. 19 is an explanatory drawing showing the status 5 . With step M 136 of FIG. 11, the management server 110 P (management module 116 P) sends resynchronization instructions to the asynchronous secondary device 200 R. The backup module 242 R which has received the resynchronization instructions copies the data of the source volume VOL 1 -P to the destination volume VOL 2 -R (backup copy), and matches the data of the destination volume VOL 2 -R to the data of the source volume VOL 1 -P (FIG. 11, R 138 ). At this time, the backup module 242 R copies only the data of the updated blocks specified by the third bitmap BMC. Also, the backup module 242 R sets the pair status of the backup copy pair BCP of the pair information 248 R to “resynchronization.” After copy completion, the backup module 242 R sets this pair status to “pair,” and furthermore, sends the resynchronization completion notification to the management server 110 P (status 5 ).

Meanwhile, it is possible that during the time from the start of this backup copy until completion, the backup volume VOL 2 -R does not have host consistency. However, the copy source volume VOL 1 -R has host consistency.

Also, when the resynchronization instructions are received, there are cases when the pair status of the second copy pair CP 2 for the pair information 248 R (FIG. 7) is not “split.” In this case, the backup module 242 R sends to the management server 110 P notification that execution is impossible for the resynchronization instructions (not illustrated) instead of backing up (FIG. 11, R 138 ). The management server 110 P (management module 116 P) repeatedly sends the resynchronization instructions to the asynchronous secondary device 200 R until resynchronization is possible.

After resynchronization completion, the system 10 a status changes to status 6 . FIG. 20 is an explanatory drawing showing the status 6 . The management server 110 P (management module 116 P) sends the split instruction to the asynchronous secondary device 200 R responsive to the resynchronization completion notification (FIG. 11, M 140 ). The backup module 242 R of the asynchronous secondary device 200 R sets the pair status of the backup copy pair BCP for the pair information 248 R (FIG. 7) to “split” (status 6 ).

After split completion, the backup module 242 R sends the split completion notification to the management server 110 P (FIG. 11, R 142 ). When this is done, the system 10 a status changes to status 7 . FIG. 21 is an explanatory drawing showing the status 7 .

Furthermore, after the split completion, the system 10 a status changes to status 8 . FIG. 22 is an explanatory drawing showing the status 8 . The management server 110 P (management module 116 P), when a split completion notification is received, sends the resynchronization instruction to the primary device 200 P (FIG. 11, M 144 ). The copy module 232 P which has received the resynchronization instruction sends to the asynchronous secondary device 200 R (FIG. 11, P 146 ) the instruction to clear the third bitmap BMC and the instruction for setting the history status of the third bitmap BMC to “undetermined.” The history creation module 238 R which received these instructions clears the third bitmap BMC and sets its history status to “undetermined” (FIG. 11, R 148 ).

Furthermore, the copy module 232 P notifies the start of resynchronization to the asynchronous secondary device 200 R (FIG. 11, P 150 ). Next, at step R 152 , the history creation module 238 R which received the notification sets the generation number of the third bitmap BMC to one newer number ( 101 ). Furthermore, the backup module 242 R which received the notification sets the pair status of the asynchronous copy pair CP 2 for the pair information 248 R (FIG. 7) to “pair.”

Next, the copy module 232 P executes resynchronization processing (P 154 ). The copy module 232 P sets the pair status of the asynchronous copy pair CP 2 for the pair information 248 P to “resynchronization.” Then, the copy module 232 P sends (P 156 ) the copy requests WR 34 R and WR 35 R to the asynchronous secondary device 200 R according to the new generation ( 101 ) requests (WR 34 , 35 ) saved in the cache memory 260 P. The asynchronous secondary device 200 R executes the write processing according to the request. Also, the history creation module 238 R updates the third bitmap BMC according to this write process. As a result, the system 10 a status becomes the status 8 shown in FIG. 22.

This status 8 is the same as the status 1 shown in FIG. 15 except for the point that the two bitmaps BMA and BMB positions are switched and the point that the generation number is updated. Hereafter, the data processing system 10 a , the same as with the procedure in FIG. 10 and FIG. 11, repeatedly executes the split and resynchronization processing.

Note that with the examples in FIG. 10 and FIG. 11, described is a case when the five host requests WR 31 to WR 35 for which write sequence changing is allowed are sent to the primary host 100 P, and even in a case when a different number of requests are sent at a different timing, the system 10 a status change is the same. Furthermore, even when a plurality of requests for which changing is not allowed are sent, the system 10 a status change is the same simply by changing the timing at which later host requests are sent.

As described above, with the second embodiment, the backup volume VOL 2 -R stores a backup of the primary device 200 P volume VOL 1 -P at the point when the primary device 200 P receives the last split instruction (FIG. 10, P 110 ). Furthermore, the third bitmap BMC is cleared before the execution of writing of the new generation copy request (FIG. 11, R 148 ). Therefore, the update history according to the new generation request is stored in the third bitmap BMC. Meanwhile, the synchronous secondary device 200 L stores the new generation bitmap and the bitmap of the generation one prior to the new generation.

B3. Secondary Copy Process

FIG. 23 is an explanatory drawing showing the data processing system 10 a when a problem occurs in the primary device 200 P. In this case as well, the same as with the first embodiment shown in FIG. 6, the secondary host 100 L restarts the data processing by using the synchronous secondary device 200 L. However, before restarting data processing, the secondary copy module 240 L executes the secondary copy processing for matching the data of the volume of the synchronous secondary device 200 L and the data of the volume of the asynchronous secondary device 200 R. The reason that this secondary copy process is executed is that after restarting of the data processing by the secondary host 100 L, the same as in normal times, the data stored in the volume VOL 1 -L is copied to the asynchronous secondary device 200 R. Here, the secondary copy module 240 L, instead of copying the whole of one volume into another volume, uses the bitmaps BMA, BMB, and BMC to specify parts between two volumes for which data is different, and copies only the data of this different part (hereafter also called “differential data”).

FIG. 24 is a flow chart showing the procedure of the secondary copy process. With the first step S 200 , the secondary copy module 240 L selects the volume having host consistency (hereafter the selected volume is also called the “consistent volume”) from among the two volumes VOL 1 -R and VOL 2 -R of the asynchronous secondary device 200 R. FIG. 25 (A) shows the condition for selecting the consistent volume. This condition is determined based on the pair status of the pair information 248 R. The secondary copy module 240 L can fetch the pair information 248 R by sending a fetch request to the backup module 242 R.

When the pair status of the asynchronous copy pair CP 2 is “split” (condition V 1 ), the secondary copy module 240 L selects the volume VOL 1 -R. Meanwhile, when the pair status of the asynchronous copy pair CP 2 is not “split,” and when the pair status of the backup copy pair BCP is “split” (condition V 2 ), the secondary copy module 240 L selects the backup volume VOL 2 -R. Note that normally, if the pair status of the asynchronous copy pair CP 2 is not “split,” then the pair status of the backup copy pair BCP is “split.”

With the next step S 210 (FIG. 24), the secondary copy module 240 L selects the bitmap for specifying the difference (also called the “differential bitmap”). Here, the differential bitmap means 1 or more bitmaps for which it is possible to specify the part for which data differs between the volume VOL 1 -L and the consistent volume. FIG. 25(B) shows the conditions for selecting the differential bitmap. The numbers in the parentheses added to the differential bitmap indicate the number of selected bitmaps. This condition is determined based on the generation number and the history status of each bitmap BMA, BMB, and BMC. Note that the secondary copy module 240 L is able to fetch information relating to the third bitmap BMC by sending the fetch request to the backup module 242 R. Details of the differential bitmap selection process are described later.

Also, following, the two bitmaps BMA and BMB of the synchronous secondary device 200 L are also called the “synchronous bitmaps BMA and BMB.” Also, the third bitmap BMC of the asynchronous secondary device 200 R is also called the “asynchronous bitmap BMC.”

With the next step S 220 (FIG. 24), the secondary copy module 240 L selects the copy direction. Normally, the secondary copy module 240 L copies the volume in which relatively new data is stored to another volume. This is because restarting of the data process is done by using relatively new host data.

Here, of the copy request generation numbers reflected in the volume, the newest number is called the “volume generation number.” The secondary copy module 240 L selects as the copy direction the direction from the volume for which the generation number is relatively new to the volume for which the generation number is relatively old. The generation number of the synchronous copy destination volume VOL 1 -L is the same as that of the newer bitmap between the synchronous bitmaps BMA and BMB. The generation number of the asynchronous copy destination volume VOL 1 -R is the same as the generation number of the asynchronous bitmap BMC. Also, when the backup volume VOL 2 -R is selected as the consistent volume, the pair status of the second copy pair CP 2 is “not split.” In this case, the generation number of the backup volume VOL 2 -R is just one older than the generation number of the asynchronous bitmap BMC.

Note that the secondary copy module 240 L may also select the copy direction according to the user instructions. The secondary copy module 240 L can fetch this kind of instruction using the operating panel (not illustrated) of the synchronous secondary device 200 L or from the management server that is able to communicate with the synchronous secondary device 200 L.

With the next step S 230 (FIG. 24), the secondary copy module 240 L performs copying of the differential data between the volume VOL 1 -L and the consistent volume, and matches the data of these volumes. With this step S 230 , the secondary copy module 240 L sends the read/write request for the consistent volume to the asynchronous secondary device 200 R. The reading and writing according to this request is executed by the read/write module 235 R.

When the differential data copying is completed, the secondary copying process ends. After that, the secondary host 100 L uses the secondary storage device 200 L (VOL 1 -L) to restart the data processing.

Next, the details of the differential bitmap selection conditions are described. Following, that bitmap within the synchronous bitmaps BMA and BMB whose generation is the same as that of the asynchronous bitmap BMC is also called simply “same generation bitmap.” Similarly, that bitmap within the synchronous bitmaps BMA and BMB whose generation is different from that of the asynchronous bitmap BMC is also called simply “different generation bitmap.”

(1) Condition B 1 :

When the status of the asynchronous bitmap BMC is “undetermined,” and the status of the different generation bitmap is “determined,” only the same generation bitmap is selected. As a status that establishes this condition B 1 , for example, there is the status 1 of FIG. 15. When a problem occurs with the primary device 200 P with this status 1 , the data of the updated block specified by the first bitmap BMA are copied from the volume VOL 1 -L to the backup volume VOL 2 -R.

When the status of the asynchronous bitmap BMC is undetermined, normally, the backup volume VOL 2 -R is selected as the consistent volume. Also, the generation of this backup volume VOL 2 -R is one older than that of the asynchronous bitmap BMC. Furthermore, when the status of the different generation bitmap is “determined”, normally, the generation of this different generation bitmap is one older than that of the asynchronous bitmap BMC. With the status 1 , the generation number of the asynchronous bitmap BMC is 100 , the generation number of the different generation bitmap (BMB) is 99 . In this way, when the condition B 1 is established, the generation of the same generation bitmap is one newer than that of the consistent volume (backup volume VOL 2 -R). Therefore, by using this same generation bitmap, it is possible to specify the part of the data that is different between the volume VOL 1 -L and the consistent volume (VOL 2 -R). Note that the same is true for the status 8 shown in FIG. 22.

(2) Condition B 2 :

When the asynchronous bitmap BMC status is “undetermined,” and the status of the different generation bitmap is “undetermined,” both of the synchronous bitmaps BMA and BMB are selected. As a status that fulfills this condition B 2 , for example, there is the status 2 shown in FIG. 16. When a problem occurs with the primary device 200 P with this status 2 , the data of the updated block shown by the bitmap selected by the OR operation of the two bitmaps BMA and BMB is copied from the volume VOL 1 -L to the backup volume VOL 2 -R.

When the asynchronous bitmap BMC is undetermined, and the different generation bitmap is undetermined, normally, the generation of this different generation bitmap is one newer than that of the asynchronous bitmap BMC. With the status 2 , the generation number of the asynchronous bitmap BMC is 100 , and the generation number of the different generation bitmap (BMB) is 101 . In this way, when the condition B 2 is established, the generation ( 100 , 101 ) of both the synchronous bitmaps BMA and BMB are newer than the generation ( 99 ) of the consistent volume (backup volume VOL 2 -R). Furthermore, the generation numbers ( 100 , 101 ) of the synchronous bitmaps are continuous with the generation number ( 99 ) of the consistent volume. Therefore, by using the bitmaps obtained by the OR operation of both the synchronous bitmaps BMA and BMB, it is possible to specify the part of the data that is different between the volume VOL 1 -L and the consistent volume (VOL 2 -R). Note that the same is true for the status 3 in FIG. 17 as well.

(3) Condition B 3 :

When the status of the asynchronous bitmap BMC is “determined,” and the status of the same generation bitmap is “determined,” that bitmap is selected from among the synchronous bitmaps BMA and BMB whose status is “undetermined” and whose generation is newer than that of the asynchronous bitmap BMC. As a status that establishes this condition B 3 , for example, there is the status 4 of FIG. 18. When a problem occurs with the primary device 200 P with this status 4 , the data of the updated block specified by the second bitmap BMB is copied from the volume VOL 1 -L to the volume VOL 1 -R.

When the status of the asynchronous bitmap BMC is determined, normally, the asynchronous copy destination volume VOL 1 -R is selected as the consistent volume. Furthermore, when the same generation bitmap status is “determined”, normally, the different generation bitmap status is “undetermined”. In this case, the generation of the different generation bitmap is one newer than that of the asynchronous bitmap BMC (specifically, the generation of the consistent volume VOL 1 -R). With the status 4 , the generation number ( 101 ) of the different generation bitmap (BMB) is one newer than the generation ( 100 ) of the asynchronous bitmap BMC. In this way, when the condition B 3 is established, the generation of that bitmap out of the synchronous bitmaps BMA and BMB whose generation is newer than that of the asynchronous bitmap BMC and whose status is “undetermined” is one newer than that of the consistent volume (VOL 1 -R). Therefore, by using this bitmap, it is possible to specify the part of the data that is different between the volume VOL 1 -L and the consistent volume (VOL 1 -R). Also, the bitmap selected here is the bitmap representing the history of the relatively new generation out of the two synchronous bitmaps BMA and BMB. Note that the same is true for the status 5 shown in FIG. 19, the status 6 shown in FIG. 20, and the status 7 shown in FIG. 21.

Note that depending on the status of the data processing system 10 a, there are cases when none of the conditions B 1 to B 3 are established. In such a case, the secondary copy module 240 L may also copy all of one volume to another volume instead of copying only the differential data.

As described above, with the second embodiment, the secondary copy module 240 L selectively uses the two synchronous bitmaps BMA and BMB according to the conditions shown in FIG. 25(B). As a result, compared to a case of always using the OR calculation results of the two synchronous bitmaps BMA and BMB, it is possible to reduce the time required for bitmap calculation and the amount of differential data to be copied. As a result, even when a problem occurs with the primary storage device 200 P, it is possible to quickly match the data of the volume VOL 1 -L of the secondary storage device 200 L and the data of the volume (consistent volume) of the secondary storage device 200 R. As a result, by using the two storage devices 200 L and 200 R, it is possible to increase the redundancy of data and to quickly restart data processing.

Also, with the second embodiment, the backup copying while the asynchronous copying is suspended and the asynchronous copying are alternately executed repeatedly. Therefore, the asynchronous secondary device 200 R is able to secure at least one volume having host consistency. As a result, even when a problem occurs with the primary device 200 P, it is possible to quickly restart the data processing by using the asynchronous secondary device 200 R.

C. Third Embodiment

FIG. 26 is an explanatory drawing showing the process that the copy module 232 P executes at the start of resynchronization of the asynchronous copy pair CP 2 for the third embodiment. There are two differences from the second embodiment described above. The first difference is that the process at the start of this resynchronization is switched according to the presence or absence of problems with the synchronous copying. The other procedures of the writing process are the same as the example shown in FIG. 10 and FIG. 11. The second difference is that a new condition is added to the conditions for selecting the differential bitmap. The constitution of the data processing system is the same as the data processing system 10 a shown in FIG. 7.

When the copy module 232 P (FIG. 7) receives the resynchronization instructions from the management server 110 P at step S 300 (FIG. 11, step M 144 ), a judgment is made of whether or not there is a problem with synchronous copying (S 305 ). The copy module 232 P can detect the presence or absence of problems using various methods. For example, when the copy module 232 P is not able to receive completion notification even when the time elapsed from sending of the copy request to the synchronous secondary device 200 L exceeds the specified time, it is judged that there is a problem. Also, when the copy module 232 P receives an error notification from the interface (not illustrated) connected to the communication path for synchronous copying, it is judged that there is a problem. The copy module 232 P is able to judge the presence or absence of a problem by comprehensively using these kinds of various conditions.

The process (step S 310 ) when it is judged that there is no problem with synchronous copying is the same as the process described with FIG. 10 and FIG. 11. As shown with step P 146 of FIG. 11, the copy module 232 P sends to the asynchronous secondary device 200 R the instruction for clearing the third bitmap BMC (also simply called “clear instruction”) and the instruction for setting the history status of the third bitmap BMC to “undetermined” (also simply called “undetermined setting instruction”). Furthermore, as shown with step P 150 , the copy module 232 P sends to the asynchronous secondary device 200 R the instruction that makes the third bitmap BMC one generation newer (the resynchronization start notification corresponds to this generation update instruction). Hereafter, the data processing system 10 a executes the data write process according to the same procedure as in FIG. 10 and FIG. 11.

Meanwhile, when it is determined that there is a problem with synchronous copying, the copy module 232 P shifts to step S 320 . In this case, the copy module 232 P sends to the asynchronous secondary device 200 R only the instruction for making the generation of the third bitmap BMC one newer without sending the clear instruction and the undetermined setting instruction.

C1. Data Write Process Details

FIG. 27 and FIG. 28 are sequence drawings showing the procedure of the data write process by the data processing system 10 a . FIG. 28 shows the latter half after the procedure of FIG. 27. The difference with the procedures shown in FIG. 10 and FIG. 11 is only that the process is switched at the time of starting resynchronization of the asynchronous copy pair CP 2 with a problem in the communication path of the synchronous copy.

With this example, even when a problem occurs with synchronous copying, the data processing continues using the primary device 200 P and the asynchronous secondary device 200 R. Also, in this case, the copy request is no longer sent to the synchronous secondary device 200 L. Furthermore, the completion notification of the host request to the primary host 100 P is sent according to receiving of the write completion notification to the source volume VOL 1 -P (for example, the completion notification C 10 a of FIG. 5). At this time, there is no consideration of whether or not the primary device 200 P receives the copy request completion notification of the asynchronous copy pair CP 2 (for example, the request C 10 e (C 10 d ) of FIG. 5).

Also, with this example, the primary host 100 P sends to the primary device 200 P the six hosts requests WRs to WRx. Here, changing of the write sequence of these requests WRs to WRx is allowed. Also, the generation number at the time that the first step P 500 is executed is 100 .

Note that each storage device 200 , the same as with the example in FIG. 5, issues the copy (write) request completion notification, but with the description below and the other embodiments described later, the illustrated and description of this are omitted. Also, the primary device 200 P executes the write processing according to the host request, but with the description below and the other embodiments described later, the illustrated and description of this are omitted.

First, the primary host 100 P sends to the primary device 200 P each request WRs to WRw in this order (not illustrated). The primary device 200 P executes the write process and the sending of the copy requests WRsL to WRwL to the synchronous secondary device 200 L according each request (steps P 500 , P 504 , P 506 , P 516 , P 520 ). However, with the example in FIG. 27, a problem occurs with the communication path of the synchronous copying, and the request WRtL and the request WRwL are not delivered to the synchronous secondary device 200 L. The synchronous secondary device 200 L executes the write processing according to the received requests WRsL, WRuL, and WRvL (L 502 , L 508 , L 518 ). For the purposes of the following description it is assumed that the communication between the primary device 200 P and the synchronous secondary device 200 L is stopped.

Meanwhile, the management server 110 P (management module 116 P) sends the split instruction to the primary device 200 P (step M 510 ). With the example in FIG. 27, the primary device 200 P receives this split instruction between the request WRu and the request WRv. The primary device 200 P which received the split instruction switches (P 512 ) the current generation number to a new number ( 101 ), sends the marker (P 514 ), and sends the write request based on the asynchronous copy to the asynchronous secondary device 200 R (P 522 ). These steps P 512 , P 514 , and P 522 are respectively the same as the steps P 112 , P 114 , and P 128 in FIG. 10. When this is done, the asynchronous secondary device 200 R receives the requests WRsR, WRtR, and WRuR for which the generation number is 100 , and executes the write processing according to these requests (R 524 ). When this is done, the status of the data processing system 10 a becomes status 21 .

FIG. 29 is an explanatory drawing showing the status 21 . In the drawing, the respective host data stored in the volume are shown using a code for which the code of the host request with the “WR” omitted. For example, the code “s” indicates the host data of the host request WRs. The same is also true for the other figures described later.

With this status 21 , the synchronous copy destination volume VOL 1 -L stores three host data s, u, and v. Also, for the first bitmap BMA, the generation number is 100 , and the history status is “undetermined.” This first bitmap BMA stores the history of the host data s and u. Meanwhile, for the second bitmap BMB, the generation number is 101 , and the history status is “undetermined.” This second bitmap BMB stores the history of the host data v.

Meanwhile, the pair status of the asynchronous copy pair CP 2 “VOL 1 -P, VOL 1 -R” is “pair.” Also, the asynchronous copy destination volume VOL 1 -R stores the three host data s, t, and u for which the generation number is 100 . For the third bitmap BMC, the generation number is 100 , and the history status is “undetermined.” This third bitmap BMC stores the history of the host data s, t, and u.

Next, the asynchronous secondary device 200 R sends to the primary device 200 P (FIG. 27, step R 525 ) the completion notification according to the copy request by asynchronous copying. When the primary device 200 P receives the completion notification of all the asynchronous copies, the primary device 200 P and the asynchronous secondary device 200 R execute the processes of making the asynchronous copy pair CP 2 “script” (P 526 , R 528 ). These steps P 526 and R 528 are respectively the same with the steps P 132 and R 134 of FIG. 10.

Next, the management server 110 P (management module 116 P) sends the resynchronization instructions and the split instructions to the asynchronous secondary device 200 R (M 530 , M 534 ). The asynchronous secondary device 200 R backs up the asynchronous copy destination volume VOL 1 -R according to these instructions (R 532 , R 536 ). These steps M 530 , R 532 , M 534 , and R 536 are respectively the same as the steps M 136 , R 138 , M 140 , and R 142 of FIG. 11.

FIG. 30 is an explanatory drawing showing the status 22 of the asynchronous secondary device 200 R executing the backup copy (resynchronization) process. With this status 22 , the pair status of the asynchronous copy pair CP 2 “VOL 1 -P, VOL 1 -R” is “split.” Also, the history status of the third bitmap BMC is “determined.”

Next, the management server 110 P (management module 116 P) sends the resynchronization instruction (M 538 ) to the primary device 200 P. With the example in FIG. 27, a problem occurs with the synchronous copy. Therefore, the copy module 232 P executes the processing according to step S 320 of FIG. 26. In specific terms, the copy module 232 P sends the resynchronization start notification to the asynchronous secondary device 200 R (P 540 ) without sending the clear instruction or the undetermined setting instruction. The succeeding steps R 542 and P 544 (FIG. 28) are respectively the same as R 152 and P 154 of FIG. 11.

Next, the management server 110 P (management module 116 P) sends the new cycle split instruction to the primary device 200 P (FIG. 28, M 550 ). The data processing system 10 a executes the new split processing (P 551 to R 560 ) according to this instruction. These steps P 551 to R 560 are respectively the same as the steps P 511 to R 528 (FIG. 27) of the previous cycle. Also, with this cycle, the primary device 200 P sends the new generation ( 101 ) requests WRvR and WRwR to the asynchronous secondary device 200 R (P 554 ). The asynchronous secondary device 200 R executes the write process according to the received request (R 556 ). When this is done, the status of the data processing system 10 a becomes the status 23 .

FIG. 31 is an explanatory drawing showing the status 23 . The backup volume VOL 2 -R stores the generation number 100 data s, t, and u. The asynchronous copy source volume VOL 1 -R stores the generation number 101 data v and w in addition to the generation number 100 data s, t, and u. The third bitmap BMC is not cleared, so in addition to the generation number 101 history (v, w), it stores the generation number 100 history (s, t, and u). Note that with the example in FIG. 31, the volume VOL 1 -P stores the new data x. This data x is the data of the even newer generation ( 102 ).

Once the split of the asynchronous copy pair CP 2 is completed, next, the management server 110 P (management module 116 P) sends the resynchronization instruction and the split instruction to the asynchronous secondary device 200 R (M 562 , M 566 ). The asynchronous secondary device 200 R backs up (R 564 , R 568 ) the asynchronous copy destination volume VOL 1 -R according to these instructions. These steps M 562 , R 564 , M 566 , and R 568 are respectively the same as the steps M 530 , R 532 , M 534 , and R 536 of FIG. 27.

FIG. 32 is an explanatory drawing showing the status 24 for which the asynchronous secondary device 200 R executes backup copy (resynchronization) processing. With this status 24 , the pair status of the asynchronous copy pair CP 2 “VOL 1 -P, VOL 1 -R” is “split.”

Next, the management server 110 P (management module 116 P) sends the new cycle resynchronization instruction to the primary device 200 P (M 570 ). The primary device 200 P and the asynchronous secondary device 200 R execute the same processes (P 572 , R 574 , P 578 ) as the processes when the resynchronization instruction is received previously (P 540 , R 542 , P 544 ).

Next, the primary device 200 P sends the write request based on the asynchronous copy to the asynchronous secondary device 200 R (P 580 ). With this step P 580 , the request WRxR for which the generation number is 102 is sent. The asynchronous secondary device 200 R executes the write process (R 582 ) according to this request WRxR. When this is done, the status of the data processing system 10 a becomes the status 25 .

FIG. 33 is an explanatory drawing showing the status 25 . With this status 25 , the generation number of the third bitmap BMC is the number 102 which is newer than that of any of the synchronous bitmaps BMA and BMB. Also, the third bitmap BMC stores the history (x) of the generation number 102 in addition to the history (s, t, u, v, w) of the generation numbers 100 and 101 .

As described above, with the third embodiment, after a problem occurs with the synchronous copy, the third bitmap BMC is not cleared, so the third bitmap BMC stores the history of all the generations from the point that the problem occurs.

C2. Secondary Copy Process

FIG. 34 is an explanatory drawing showing the conditions for selecting the differential bitmap. The difference from the embodiment shown in FIG. 25(B) is only that the conditions B 4 and B 5 are added. When a problem occurs with the primary device 200 P (FIG. 7), the secondary copy module 240 L executes secondary copy processing the same as with the second embodiment shown in FIG. 24. However, the selection of the differential bitmap (S 210 ) is performed according to the condition of FIG. 34 instead of the condition in FIG. 25(B).

With the third embodiment, even when a problem occurs with the synchronous copy, the data processing continues using the primary 200 P and the asynchronous secondary device 200 R. As a result, it is possible for the generation number of the asynchronous bitmap BMC to be newer than the generation number of any of the synchronous bitmaps BMA and BMB. Therefore, to each of the conditions B 1 to B 3 described above, is added the condition that, “the generation number of one of the synchronous bitmaps BMA and BMB be the same as the generation number of the asynchronous bitmap BMC.” However, the status of the data processing system 10 a for which each condition has been established is the same as for the second embodiment described above. Also, the new conditions B 4 and B 5 indicate the status that can occur when a problem occurs with the synchronous copy.

(4) Condition B 4 :

When the status of the asynchronous bitmap BMC is “determined” and the status of the same generation bitmap is “undetermined,” in addition to the asynchronous bitmap BMC, the undetermined bitmap within the synchronous bitmaps BMA and BMB is selected. As a status that establishes this condition B 4 , there is the status 22 of FIG. 30, for example. With this status 22 , when a problem occurs with the primary device 200 P, the three bitmaps BMA, BMB, and BMC are selected. Then, the data of the updated blocks shown by the bitmaps obtained by the OR operation of the three bitmaps BMA, BMB, and BMC are copied between the volume VOL 1 -L and the volume VOL 1 -R. In this case, the respective two volumes store data that are not mutually shared (data t and data v). Because of this, even when priority is given to relatively new data, the secondary copy module 240 L cannot determine a desirable copy direction. In this kind of case, it is desirable for the secondary copy module 240 L to select the copy direction according to the user instruction.

Normally, the undetermined synchronous bitmaps BMA and BMB have many cases of including update histories not stored in the consistent volume of the asynchronous secondary device 200 R. Meanwhile, the asynchronous bitmap BMC continues to store history without being cleared in cases when a problem occurs with the synchronous copy. Therefore, it is possible that the asynchronous bitmap BMC will contain update histories that are not stored in the synchronous copy destination volume VOL 1 -L. Therefore, when the condition B 4 is established (when none of the three previously described conditions B 1 , B 2 , and B 3 are established), it is preferable that the undetermined bitmap within the synchronous bitmaps BMA and BMB and the asynchronous bitmap BMC be selected. Here, if the secondary copy module specifies differential data using the bitmap obtained by the OR operation of the selected bitmaps, it is possible to execute differential data copying without omission. Note that this is also the same for the status 23 shown in FIG. 31 and the status 24 shown in FIG. 32.

(5) Condition B 5 :

When the generation number of the asynchronous bitmap BMC is newer than the generation number of any of the synchronous bitmaps BMA and BMB, then only the asynchronous bitmap BMC is selected. As a status that establishes this condition B 5 , there is the status 25 of FIG. 33, for example. With this status, when a problem occurs with the primary device 200 P, the data of updated blocks shown by the asynchronous bitmap BMC is copied from the backup volume VOL 2 -R to the volume VOL 1 -L.

When the condition B 5 is established, even after a problem occurs with the synchronous copy, there are cases when the data processing continues using the asynchronous copy. In this kind of case, the newest data is stored in the consistent volume of the asynchronous secondary device 200 R. Also, the asynchronous bitmap BMC stores all the history from after the generation at the point that the problem occurs with the synchronous copy. Therefore, by using the asynchronous bitmap BMC, it is possible to specify the part of the data that is different between the volume VOL 1 -L and the consistent volume.

Note that the asynchronous bitmap BMC stores the history of the asynchronous copy destination volume VOL 1 -R instead of the backup volume VOL 2 -R. However, the asynchronous bitmap BMC includes the history of updates already reflected in the backup volume VOL 2 -R. Therefore, even when the backup volume VOL 2 -R is selected as the consistent volume, by using the asynchronous bitmap BMC, it is possible to specify differential data without omission.

As described above, with the third embodiment, by using the asynchronous bitmap BMC for storing the updated history by the asynchronous copy, the secondary copy module 240 L is able to specify the differential data (FIG. 34, conditions B 4 , B 5 ). Therefore, even when a problem occurs with the primary device 200 P after a problem occurred with the synchronous copy before that, by copying only the differential data, it is possible to quickly match the data of the volume VOL 1 -L of the secondary storage device 200 L and the data of the volume (consistency volume) of the secondary storage device 200 R.

D. Fourth Embodiment

FIG. 35 is a flow chart showing the procedure of the synchronous copy problem handling process executed by the primary device 200 P for the fourth embodiment. In contrast to the third embodiment described above, when a problem occurs with the synchronous copy, the primary device 200 P executes the split processing of the asynchronous copy pair CP 2 spontaneously even when there is no split instruction from the management server 110 P. The other procedures of the write process are the same as the examples shown in FIG. 26, FIG. 27, and FIG. 28. Also, the constitution of the data processing system is the same as the data processing system 10 a of the FIG. 7.

The copy module 232 P repeatedly executes the process of judging whether or not there is a problem with the synchronous copy (S 500 ). When there is a problem, with the next step S 510 , the status of the asynchronous copy pair CP 2 is confirmed. When the pair status is already “split,” or when the “split instruction” is already received and splitting starts, the copy module 232 P returns to the normal process without executing a spontaneous split process.

Next, the copy module 232 P shifts to step S 520 , and starts the split process spontaneously. This spontaneous split process is the same as the split process executed according to instructions from the management server 110 P. The generation setting module 236 P receives instructions from the copy module 232 P and updates the generation number (step S 525 ). Thereafter, the same processes as in FIG. 26, FIG. 27, and FIG. 28 continue to be executed.

Also, the copy module 232 P detects the host request for which the synchronous copy could not be completed at step S 500 . Furthermore the copy module 232 P handles this host request as a new generation request after the spontaneous split when the spontaneous split is executed at step S 520 .

Note that the copy module 232 P executes the spontaneous split process before the process of sending to the asynchronous secondary device 200 R the copy request according to the host request for which a synchronous copy completion notification could not be received due to a problem. This is for the asynchronous secondary device 200 R to secure a volume backup when a problem occurs. As a result, with the fourth embodiment, it is easy to restart the data process from the point that the problem occurs.

E. Fifth Embodiment

FIG. 36 is a schematic drawing showing the constitution of the data processing system 10 b for the fifth embodiment. There are three points of difference with the data processing system 10 a shown in FIG. 7. The first difference is that at the asynchronous secondary device 200 R, the history creation module 238 R and the third bitmap BMC are omitted, and instead of these, provided is a volume VOL 3 -R for storing the status information 280 R. The second difference is that the information relating to the second copy pair CP 2 for the pair information 248 R is omitted. The third difference is that the conditions used with the secondary copy process are different from the example shown in FIGS. 25(A) and (B). The other constitution is the same as the data processing system 10 a of FIG. 7. Also, the procedure of the data write process by the data processing system 10 b is the same as the procedure shown in FIG. 10 and FIG. 11 excluding the points below. Specifically, with the fifth embodiment, the process relating to setting the asynchronous bitmap BMC and the process relating to the setting of the asynchronous copy pair CP 2 of the pair information 248 R are omitted. Instead of these, the status information 280 R write process is added.

E1. Data Write Process

FIG. 37 is a sequence drawing showing the procedure of the write process of the status information 280 R. This sequence drawing shows part of the data write process by the data processing system 10 b , and shows the process that continues after the asynchronous copy is completed. The copy module 232 P, after receiving completion notification of all the copy requests of the asynchronous copying, sets the pair status of the asynchronous copy pair CP 2 for the pair information 248 P to “split” (P 600 ). This step P 600 is the same as the step P 132 of FIG. 10 (however, the completion notification is not sent). Next, the copy module 232 P sends the write request for the volume VOL 3 -R to the asynchronous secondary device 200 R (P 602 ). The write data is data indicating “determined.” The meaning of “determined” is described later. The read/write module 235 R of the asynchronous secondary device 200 R writes to the VOL 3 -R the data requested according to the received request (R 604 ). The data written here corresponds to the status information 280 R. Thereafter, the data processing system 10 b executes the same processes as those in FIG. 10 and FIG. 11 until the management server 110 P sends the resynchronization instruction of the asynchronous copy pair CP 2 to the primary device 200 P (e.g. FIG. 11, M 144 ).

FIG. 38 is a sequence drawing showing the procedure of the writing process of the status information 280 R. This sequence drawing shows part of the data writing process according to the data process system 10 b and shows the process executed ahead of the execution of the resynchronization process of the asynchronous copy pair CP 2 . First, the copy module 232 P receives resynchronization instructions from the management server 110 P (P 610 ). Next, before executing the resynchronization process, the copy module 232 P sends the write request for the volume VOL 3 -R to the asynchronous secondary device 200 R (P 612 ). The requested write data is data indicating “undetermined.” The meaning of “undetermined” is described later. The status information 280 R is updated by this data. Next, the copy module 232 P executes the resynchronization process (P 616 ) with receiving of the completion notification (not illustrated) of write requests of the status information 280 R from the read/write module 235 R. This step P 616 is the same as the step P 154 of FIG. 11. Thereafter, the data processing system 10 b executes the same processes as in FIG. 10 and FIG. 11 until the management server 110 P sends the split instruction to the primary device 200 P (e.g. FIG. 10, M 108 ).

As described above, with the fifth embodiment, the status information 280 R alternately indicates “determined” and “undetermined” according to the pair status of the asynchronous copy pair CP 2 . Here, the status information 280 R is “determined” during the period from the time the asynchronous copying relating to one generation is completed until the time the asynchronous copying of the next generation starts. Specifically, the fact that the status information 280 R is “determined” means that the asynchronous copy destination volume VOL 1 -R has host consistency. Meanwhile, the fact that the status information 280 R is “undetermined” means that it is possible that the destination volume VOL 1 -R does not have host consistency. Note that in this case, normally, the backup volume VOL 2 -R has host consistency.

FIG. 39 is a sequence drawing showing the procedure of the status information 280 R write process. This sequence drawing shows part of the data write process by the data processing system 10 b , and shows the process executed when a problem occurs with the synchronous copy. When a problem occurs with the synchronous copy, the copy module 232 P sends to the asynchronous secondary device 200 R copy requests according to the host requests for which it is not possible to receive the completion notification of the synchronous copy. Before sending this kind of copy request to the asynchronous secondary device 200 R, the copy module 232 P sends to the asynchronous secondary device 200 R (P 620 ) the write request of data showing “copy all.” The meaning of “copy all” is described later. The read/write module 235 R writes the requested data to the VOL 3 -R (R 622 ) according to the received request. Note that the status information 280 R includes data indicating “copy all” in addition to data indicating “determined/undetermined.”

Next, the copy module 232 P sends to the asynchronous secondary device 200 R (P 624 ) the copy (write) request according to the asynchronous copy responsive to receiving of the completion notification (not illustrated) of the write requests of the status information 280 R from the read/write module 235 R. Included in this copy request are copy requests of host data for which synchronous copying is not done. Note that this step P 624 is performed the same as step P 128 of FIG. 10 and step P 156 of FIG. 11. Thereafter, the data processing system 10 b continues the data writing process the same as the examples in FIG. 10 and FIG. 11.

E2. Secondary Copy Process

With the fifth embodiment, when a problem occurs in the primary device 200 P, the secondary copy module 240 L executes the secondary copy process according to the procedure shown in FIG. 24. FIG. 40(A) shows the conditions for selecting the consistent volume. The difference with the conditions shown in FIG. 25(A) is that instead of the pair status of the second copy pair CP 2 , the status information 280 R is used. As described above, when the status information 280 R indicates “determined” (condition V 11 ), the volume VOL 1 -R has host consistence, so the volume VOL 1 -R is selected. Meanwhile, when the status information 280 R indicates “undetermined,” and when the pair status of the backup copy pair BCP is “split” (condition V 12 ), the backup volume VOL 2 -R is selected. Note that normally, when the status information 280 R indicates “undetermined,” the pair status of the backup copy pair BCP is “split.” Also, the secondary copy module 240 L is able to fetch the status information 280 R by communication with the read/write module 235 R.

FIG. 40(B) shows the condition for selecting the differential bitmap. The difference with the condition shown in FIG. 25(B) is that this condition is determined using the status information 280 R.

(1) Condition B 11 :

When the status information 280 R indicates “copy all,” the secondary copy module 240 L, instead of a differential copy, copies all of one volume to another volume. In this case, the asynchronous copy destination volume VOL 1 -R stores the data not stored in the synchronous copy destination volume VOL 1 -L. However, with the fifth embodiment, the asynchronous bitmap BMC is omitted, so specification of this differential data is not possible. In light of this, the secondary copy module 240 L executes copying of the entire volume.

(2) Condition B 12 :

When the status information 280 R does not indicate “copy all,” and does indicate “determined,” only one undetermined bitmap within the synchronous bitmaps BMA and BMB is selected. When this condition B 12 is established, normally, of the two synchronous bitmaps BMA and BMB, the bitmap of the same generation as the consistent volume VOL 1 -R is determined. Meanwhile, the other synchronous bitmap is undetermined, and furthermore, stores a history of one newer generation than that of the consistent volume VOL 1 -R. Therefore, the secondary copy module 240 L is able to specify the differential data by using this undetermined synchronous bitmap.

(3) Condition B 13 :

When the status information 280 R does not indicate “copy all” and does indicate “undetermined,” both the synchronous bitmaps BMA and BMB are selected. When this condition B 13 is established, normally, the same as with the condition B 2 shown in FIG. 25(B), the generation of both the synchronous bitmaps BMA and BMB is newer than the generation of the consistent volume (backup volume VOL 2 -R). Therefore, the secondary copy module 240 L is able to specify the differential data using the bitmap obtained by the OR operation of the two bitmaps BMA and BMB.

As described above, with the fifth embodiment, in contrast to the example shown in FIG. 7, the third bitmap BMC is omitted. Furthermore, to store the information relating to the status of the asynchronous copy destination volume VOL 1 -R in the asynchronous secondary device 200 R, the copy module 232 P sends to the asynchronous secondary device 200 R write requests including simple data (status information 280 R) instead of various instructions (for example, notification that indicates request to set the pair status of the step P 150 of FIG. 11). Therefore, while trying to simplify the functions of the asynchronous secondary device 200 R, it is possible to do secondary copy processing using the volume having host consistency.

F. Sixth Embodiment

FIG. 41 is an explanatory drawing showing the constitution of the data processing system 10 c for the sixth embodiment. The difference from the embodiments described above is that consistency groups CG 1 and CG 2 are set. Here, the consistency groups are groups of a plurality of copy pairs for which data write processing control is managed with a common generation. With this embodiment, the shift to split status is performed simultaneously for a plurality of copy pairs included in one consistency group. To say this another way, generation switching is performed simultaneously for this plurality of copy pairs. Also, a common generation number is used for the plurality of copy pairs of one consistency group.

The memory 114 P of the management server 110 P stores the management module 116 a P and the consistency group information 118 . The consistency group information 118 determines the correlation between the asynchronous copy source volume identifier and the group number. With the example in FIG. 41, two volumes VOL 10 -P and VOL 11 -P (two asynchronous copy pairs ACP 10 and ACP 11 ) constitute first group CG 1 and the three volumes VOL 12 -P, VOL 13 -P, and VOL 14 -P (three asynchronous copy pairs ACP 12 , ACP 13 , and ACP 14 ) constitute the second group CG 2 . Note that the plurality of source volumes VOL 10 -P to VOL 14 -P constituting the copy pairs ACP 10 to ACP 14 are specified by mutually different identifiers. The same is also true for the plurality of destination volumes VOL 20 -R to VOL 24 -R. Also, the other constitution of the data processing system 10 c is the same (not illustrated) as the embodiments described above (e.g., the data processing system 10 a in FIG. 7 and the data processing system 10 b in FIG. 36).

FIG. 42 is a sequence drawing showing a comparative example of data write processing. This sequence drawing shows part of the data write process by the data processing system 10 c , and shows the part executed by the split of the two asynchronous copy pairs ACP 10 and ACP 11 . With this comparative example, the splitting of the two asynchronous copy pairs ACP 10 and ACP 11 is executed at different times. Note that the processing other than the split is the same as the embodiments described above (e.g. the examples in FIG. 10 and FIG. 11 or in FIG. 26, FIG. 27, and FIG. 28). Furthermore, it is assumed that the volume generation is switched from 100 to 101 by this split.

With the example in FIG. 42, the primary device 200 P receives the four host requests WRh to WRk in that sequence, and sends to the asynchronous secondary device 200 R the asynchronous copy requests WRhR to WRkR according to each request (P 200 , P 202 , P 204 , and P 206 ). The asynchronous secondary device 200 R executes the write processing according to the requests (R 201 , R 203 , R 205 , and R 207 ).

Here, the subject of the first two copy requests WRhR and WRiR is the volume VOL 20 -R (asynchronous copy pair ACP 10 ), and the subject of the latter two copy requests WRjR and WRkR is the volume VOL 21 -R (asynchronous copy pair ACP 11 ). The reason that the subject volumes are different is that the subject of the first two host requests WRh and WRi is the volume VOL 10 -P (FIG. 41), and the subject of the latter two host requests WRj and WRk is the volume VOL 11 -P. Furthermore, it is assumed that changing of the write sequence of these host requests WRh to WRk is prohibited. The prohibition of the write sequence occurs, for example, when a plurality of host requests are destined to a plurality of volumes which constitute a single database, or when a plurality of host requests are destined to a plurality of volumes which constitute a single data file.

The management module 116 a P (FIG. 41) references the consistency group information 118 , and sends to the primary device 200 P the split instructions of all the asynchronous copy pairs (ACP 10 , ACP 11 ) included in the first consistency group CG 1 . However, with this comparative example, the split instructions of each copy pair are sent at different times to each other (M 210 , M 212 ). The primary device 200 P executes the split (P 210 ) of the asynchronous copy pair ACP 10 between steps P 200 and P 202 . Furthermore, the primary device 200 P executes the split (P 212 ) of the asynchronous copy pair ACP 11 between steps P 204 and P 206 . The generation switches from 100 to 101 with these splits. The numbers in the parentheses added to the write processes of FIG. 42 (R 201 , R 203 , R 205 , R 207 ) indicate the generation numbers.

Here, when a problem occurs with the primary device 200 P, by using the asynchronous secondary device 200 R, data processing restarts. However, in the two volumes VOL 20 -R and VOL 21 -R (or the backup volumes of these) in which all the requests of generation number 100 are reflected do not have the prior host data i stored, but rather have the latter host data j stored. Specifically, the whole of these volumes VOL 20 -R and VOL 21 -R (or the backup volumes of these) do not have host consistency. As a result, a problem occurs with restarting of the data processing using the asynchronous secondary device 200 R.

FIG. 43 is a sequence drawing showing the data write process for the sixth embodiment. The difference from the comparative example of FIG. 42 is that the splits of the two asynchronous copy pairs ACP 10 and ACP 11 are performed at the same time. The data write process executed for each of the copy pairs ACP 10 and ACP 11 is the same as each of the embodiments described above (e.g. the example in FIG. 10 and FIG. 11 or in FIG. 26, FIG., 27 , and FIG. 28) for the part other than the split timing. The management module 116 a P (FIG. 41) sends (M 220 ) to the primary device 200 P instructions to simultaneously execute the split of the first consistency group CG 1 (asynchronous copy pairs ACP 10 and ACP 11 ). The primary device 200 P executes splitting simultaneously (P 220 ) according to the instructions. With the example in FIG. 43, the primary device 200 P executes the split between steps P 204 and P 206 . As a result, the two volumes VL 20 -R and VOL 21 -R (or the backup volumes of these) for which all the requests of generation number 100 are reflected store the host data h, i, and j. The entirety of these volumes VOL 20 -R and VOL 21 -R (or the backup volumes of these) have host consistency. This is also the same in cases when the split is performed at another time. This is also the same when the subject volume of each host request is different from the example in FIG. 43.

As described above, with the fifth embodiment, the management module 116 a P sends to the primary device 200 P the instruction to split simultaneously for the plurality of copy pairs included in one consistency group. The primary device 200 P executes splitting (generation updating) of these plurality of pairs at the same time according to this instruction. As a result, even when a plurality of host data for which changing of the write sequence is not allowed are stored divided into a plurality of volumes, for the entirety of the volumes or the entirety of the backup volumes after splitting, it is possible to prevent the host data write sequence from being changed across the generation. This is also the same when the number of copy pairs is three or more included in one consistency group.

G. Seventh Embodiment

FIG. 44 is an explanatory drawing showing the constitution of the data processing system 10 d for the seventh embodiment. The difference from the data processing system 10 c shown in FIG. 41 is that the plurality of asynchronous copy pairs of one consistency group is provided divided into the plurality of primary devices 200 P and asynchronous secondary devices 200 R. This data processing system 10 d has a plurality of primary devices 200 a P to 200 c P and a plurality of asynchronous secondary devices 200 a R to 200 c R. Regarding the first consistency group CG 1 , the first asynchronous copy pair ACP 10 is formed using the first primary device 200 a P and the first asynchronous secondary device 200 a R, and the second asynchronous copy pair ACP 11 is formed using the second primary device 200 b and the second asynchronous secondary device 200 b R. For the second consistency group CG 2 , similarly, the asynchronous copy pair ACP 12 is formed using the second primary device 200 b P and the second asynchronous secondary device 200 b R, and the asynchronous copy pair ACP 13 and ACP 14 are formed using the third primary device 200 c P and the third asynchronous secondary device 200 c R. Note that the constitution of each primary device 200 a P to 200 c P is the same as the constitution of the primary device 200 P shown in FIG. 7, and the constitution of each asynchronous secondary device 200 a R to 200 c R is the same as the constitution of the asynchronous secondary device 200 R shown in FIG. 7. The other constitution of the data processing system 10 d is the same as that of the data processing system 10 a of FIG. 7 (illustration omitted).

FIG. 45 is a sequence drawing showing a comparative example of the data write process. This sequence drawing shows part of the data write process by the data processing system 10 d , and shows the part for which the split of the two asynchronous copy pairs ACP 10 and ACP 11 is executed. Note that the data write process executed for each copy pair ACP 10 and ACP 11 is the same as the example in FIG. 10 and FIG. 11. Furthermore, it is assumed that the generation of the volume is switched from 100 to 101 by this split.

The primary host 100 P sends the four hosts requests WRh to WRk that are the same as the example in FIG. 42 in this sequence (H 300 , H 310 , H 320 , and H 330 ). However, the sending destination of the prior two requests WRh and WRi is the first primary device 200 a P, and the sending destination of the latter two requests WRj and WRk is the second primary device 200 b P. Each primary device 200 a P and 200 b P execute the write process (Pa 302 , Pa 312 , Pb 322 , Pb 332 ) according to the received requests, and sends to the primary host 100 P a completion notification (Pa 304 , Pa 314 , Pb 324 , and Pb 334 ).

The management module 116 b P (FIG. 44) sends the split instructions (M 340 ) simultaneously to the two primary device 200 a P and 200 b P (control device 210 a P and 210 b P). However, each primary device 200 a P and 200 b P receive split instructions at mutually different times. This is because the communication path with the management server 110 P is different for each primary device 200 a P and 200 b P. As a result, the splits of each asynchronous copy pair ACP 10 and ACP 11 are executed at mutually different times. With the example of FIG. 45, the first primary device 200 a P executes the split between steps Pa 304 and Pa 312 . Meanwhile, the second primary device 200 b P executes the split between steps Pb 324 and Pb 332 . The number in the parentheses attached to the write process (Pa 302 , Pa 312 , Pb 322 , Pb 332 ) of FIG. 45 indicates the generation number.

Here, the data processing is restarted by using the asynchronous secondary devices 200 a R and 200 b R. However, the same as with the comparative example in FIG. 42, the entirety of the two volumes VOL 20 -R and VOL 21 -R (or the backup volumes) does not have host consistency, so a problem occurs with restarting of the data process using the asynchronous secondary devices 200 a R and 200 b R.

FIG. 46 is a sequence drawing showing the data write process for the seventh embodiment. The difference with the comparative example of FIG. 45 is that the management module 116 b P (FIG. 44) sends split instructions to the two primary devices 200 a P and 200 b P during the time in which the data update is postponed by the primary host 100 P. The four host requests WRh to WRk are the same as the example in FIG. 45. In specific terms, first, the management server 110 P (management module 116 b P) sends an instruction for postponing the new data update to the primary host 100 P (M 350 ) before sending the split instruction. When this is done, the primary host 100 P sends an acknowledgement of the instruction for postponing to the management server 110 P (H 352 ) after receiving from the primary devices 200 a P, 200 b P the completion notifications of all the host requests already sent. With the example in FIG. 46, the primary host 100 P receives the postponement instruction after sending of the host request WRh (H 300 ). Therefore, the primary host 100 P sends the acknowledgement to the management server 110 P (H 352 ) after receiving the completion notification Ch of this request WRh. Thereafter, the primary host 100 P postpones the sending of the new host request.

Next, the management server 110 P (management module 116 b P) sends split instructions (M 354 ) to the two primary devices 200 a P and 200 b P responsive to receiving of the acknowledgement of the instruction for postponing from the primary host 100 P. By doing this, each primary device 200 a P and 200 b P respectively executes the split according to instructions, and sends the split completion notification to the management server 110 P (Pa 356 , Pb 358 ). The copy module 232 P of each primary device 200 a P and 200 b P, respectively, sends these completion notifications responsive to the generation update and to the receiving of the completion notification of all the copy requests of the asynchronous copy of the generation before the update.

The management server 110 P sends to the primary host 100 P (M 360 ) instructions for restarting the new data update responsive to receiving the split completion notification from the two primary devices 200 a P and 200 b P. The primary host 100 P sends the next host request WRi (H 310 ) responsive to receiving this instruction. Thereafter, the data processing system 10 d executes processing according to each host request WRi, WRj, and WRk (the illustration is omitted for the requests WRj and WRk).

As described above, with the seventh embodiment, the management module 116 b P sends instructions for postponing the new data update to the primary host 100 P. Then, the management module 116 b P sends the split instructions for all the copy pairs included in one consistency group to all the primary storage devices ( 200 a P and 200 b P, in specific terms, all the control devices ( 210 a P and 210 b P)) which control the pair status of each copy pair during postponing of the data update by the primary host 100 P. As a result, even in cases when a plurality of host data for in-order write are stored in a plurality of volumes whose pair status are controlled by different control devices respectively, it is possible to prevent the host data write sequence from being changed across the generation for the whole of the volumes or the whole of the backup volumes after splitting (here, the plurality of host data for in-order write means the plurality of host data for which changing of the write sequence is not allowed). This is also the same in cases when the number of those primary storage devices (control devices) is three or more which controls each pair status of the plurality of copy pairs contained in one consistency group. Also, this aspect of this embodiment can also be applied to the example in FIG. 26, FIG. 27, and FIG. 28 or in FIG. 35 and FIG. 36.

H. Eighth Embodiment

FIG. 47 is an explanatory drawing showing the constitution of the data processing system 10 e for the eighth embodiment. The difference with the data processing system 10 d shown in FIG. 44 is only that the management module 116 c P sends the update postponement instructions to each primary device 200 P instead of to the primary host 100 P. The other constitution is the same as that of the data processing system 10 d shown in FIG. 44.

FIG. 48 is a sequence drawing showing the data write process for the eighth embodiment. The difference from the example shown in FIG. 46 is that the split instructions are sent to each primary device 200 a P and 200 b P during postponement of the data update by each primary device 200 a P and 200 b P. The four host requests WRh to WRk (H 300 to H 330 ) and the processes (Pa 302 to Pa 314 and Pb 322 to Pb 334 ) of each primary device 200 a P and 200 b P according to these requests are the same as those in the example in FIG. 45 and FIG. 46.

First, the management server 110 P (management module 116 c P) sends instructions for postponing new data updates to the two primary devices 200 a P and 200 b P (M 370 ). Each of the primary devices 200 a P and 200 b P sends an acknowledgement of the instructions for postponing to the management server 110 P (Pa 372 , Pb 374 ) after issuing a completion notification of all the host requests which have already been received. With the example in FIG. 48, the first primary device 200 a P receives the postponement instructions after receiving the host request WRh (Pa 302 ). Therefore, the first primary device 200 a P sends the acknowledgement to the management server 110 P (Pa 372 ) after sending the completion notification Ch of this request WRh (Pa 304 ). Thereafter, each primary device 200 a P and 200 b P postpones the write process without executing the write process even when a new host request is received. For example, with the example in FIG. 48, the first primary device 200 a P receives the host request WRi after sending of the acknowledgement (Pa 372 ). However, the first primary device 200 a P executes the process according to this request later (described later).

Next, the management server 110 P (management module 116 c P) sends split instructions (M 380 ) to each primary device 200 a P and 200 b P responsive to receiving the acknowledgement of the instructions for postponing from each primary device 200 a P and 200 b P. By dong this, each primary device 200 a P and 200 b P, the same as with the example in FIG. 46, executes the splitting according to the respective instructions, and sends the split completion notification to the management server 110 P (Pa 382 , Pb 384 ). The management server 110 P sends the instruction to restart new data updating to each primary device 200 a P and 200 b P (M 390 ) responsive to receiving the split completion notification from the two primary devices 200 a P and 200 b P. Each primary device 200 a P and 200 b P starts execution of processing according to the new host request responsive to receiving this instruction. For example, with the example in FIG. 48, the first primary device 200 a P executes the write process (Pa 312 ) according to the postponed request WRi responsive to receiving the restart instruction. Thereafter, the data processing system 10 e executes processing according to the requests WRj and WRk.

As described above, with the eighth embodiment, the management module 116 c P sends instructions for postponing the new data update to all the primary storage devices ( 200 a P and 200 b P, in specific terms, all the control devices ( 210 a P and 210 b P)) for controlling the pair status of each copy pair of all the copy pairs included in one consistency group. Then, the management module 116 c P sends to all of these primary storage devices the split instructions for all the copy pairs included in one consistency group during postponement of data update by all of these primary devices. As a result, even in cases when a plurality of host data for in-order write are stored in a plurality of volumes whose pair status are controlled by different control devices respectively, it is possible to prevent the host data write sequence from changing across the generation for the whole of the volumes or the whole of the backup volumes after splitting. This is also the same in cases when the number of those primary storage devices (control devices) is three or more which controls each pair status of the plurality of copy pairs contained in one consistency group. Also, this aspect of this embodiment can also be applied to the example in FIG. 26, FIG. 27, and FIG. 28 or in FIG. 35 and FIG. 36.

I. Variation Examples

Note that within the structural elements for each of the embodiments noted above, the elements other than the elements claimed in independent claims are additional elements, and can be suitably omitted. Also, this invention is not limited to the embodiments or aspects noted above, but can be implemented with various aspects within a scope that does not stray from its key points, and variations such as the following are possible, for example.

Variation Example 1

With the embodiments noted above, it is possible to use various conditions as the condition for selecting the bitmap to be used with the secondary copy processing. For example, instead of the condition shown in FIG. 25(B), it is possible to use any conditions for selecting as the differential bitmap a bitmap of a newer generation than the generation of the consistency volume within the two synchronous bitmaps BMA and BMB. Note that with the example in FIG. 25(B), this kind of differential bitmap is selected based on the generation of the asynchronous bitmap BMC. Therefore, the generation of the asynchronous bitmap BMC corresponds to the “backup generation information” for the present invention.

Here, the secondary copy module 240 L (FIG. 7) may also fetch the generation of the backup volume VOL 2 -R from the backup module 242 R. As a method for the backup module 242 R to specify the generation of the backup volume VOL 2 -R, various methods can be used. For example, it is possible to use a method of specifying based on the asynchronous bitmap BMC generation and on the progress status of the backup process. In specific terms, before the update recorded in the asynchronous bitmap BMC is reflected in the backup volume VOL 2 -R by the backup process, the generation of the backup volume VOL 2 -R is the generation one prior to the generation of the asynchronous bitmap BMC. Meanwhile, after reflection, the generation of the backup volume VOL 2 -R is the same as the generation of the asynchronous bitmap BMC. Also, the same as with the copy (write) request shown in FIG. 12, the primary device 200 P may also send a copy request to which the generation number GNo is added to the asynchronous secondary device 200 R. In this way, the backup module 242 R is able to execute specification based on the generation number GNo contained in the copy request and the progress status of the backup process.

Variation Example 2

For the condition shown in FIG. 25(A), the fact that the pair status of the asynchronous copy pair CP 2 (VOL 1 -R) is “split” means that there is establishment of the first timing condition representing that the point of execution of the secondary copy process is during the time from after the completion notification is issued for all the asynchronous copy requests of the old generation until starting of sending the asynchronous copy requests of the new generation. Also, the fact that the pair status of the backup copy pair BCP (VOL 2 -R) is “split” means that there is establishment of the second timing condition representing that the point of execution of the secondary copy process is during the time from after the backup completion notification is issued until the start of the new backup process. In this way, the pair status of the asynchronous copy pair CP 2 and the pair status of the backup copy pair BCP correspond to the “timing condition information.” However, with each of the embodiments noted above, normally, when the pair status of the asynchronous copy pair CP 2 (VOL 1 -R) is “not split,” the pair status of the backup copy pair BCP (VOL 2 -R) is “split.” Therefore, the secondary copy module 240 L is able to judge that the second timing condition is established according to the fact that the pair status of the asynchronous copy pair CP 2 is “not split” without using the pair status of the backup copy pair BCP.

For the conditions shown in FIG. 40(A), the fact that the status information 280 R indicated “determined” means that there is establishment of the first timing condition representing that the point of execution of the secondary copy process is during the time from after the completion notification is issued for all the synchronous copy requests of the old generation until the start of sending the asynchronous copy requests of the new generation. Also, with the embodiments noted above, normally, when the status information 280 R does not indicate “determined,” specifically, when the status information 280 R indicates “undetermined,” the pair status of the backup copy pair BCP is “split.” Therefore, the secondary copy module 240 L is able to judge that the second timing condition is established according to the fact that the status information 280 R is “undetermined” without using the pair status of the backup copy pair BCP (VOL 2 -R).

Variation Example 3

For the embodiments shown in FIG. 41, FIG. 44, and FIG. 47, it is possible to use various constitutions as the constitution of the synchronous copy destination volume. For example, the plurality of synchronous copy destination volumes (not illustrated) respectively correlated to the source volumes VOL 10 -P to VOL 14 -P may also be provided to one unit of the synchronous secondary device 200 L. Instead of this, it is also possible for the plurality of destination volumes to be provided divided into a plurality of storage devices. In any case, it is preferable that the respective plurality of destination volumes be specified by mutually different identifiers (device identifiers and volume identifiers). The same is also true for the constitution of the asynchronous copy destination volumes.

Variation Example 4

With each of the embodiments noted above, as a device for sending instructions to a storage device 200 according to the operating status of any of the other storage devices 200 , this is not limited to the management server 110 P, and it is possible to use various devices. For example, it is possible to have each of the storage devices 200 mutually send instructions to the other storage devices 200 according to its own operation status. In specific terms, the primary device 200 P may send the resynchronization (backup) instructions (FIG. 11, M 136 ) to the asynchronous secondary device 200 R responsive to receiving the completion notification of all the old generation copy requests based on the asynchronous copy (e.g. FIG. 10, R 131 ) from the asynchronous secondary device 200 R. Furthermore, responsive to the completion of the backup process (FIG. 11, R 138 ), the asynchronous secondary device 200 R may send the resynchronization instructions (FIG. 11, M 144 ) to the primary device 200 P. Also, the control device 210 of any of the storage devices 200 may have the function of the management server 110 P.

Variation Example 5

For each of the embodiments noted above, the synchronous secondary device 200 L (e.g. the history creation module 238 L, FIG. 7) may use various methods as the method of judging whether or not the completion notification has been issued for all the synchronous copy request of a certain generation. For example, it is possible to have the copy module 232 P send to the synchronous secondary device 200 L the total number of host request of a certain generation. The history creation module 238 L is able to judged that all the completion notification of that generation have been issued when the total number of the completion notification of the copy requests of that generation is the same as the total number of host requests of the same generation.

Similarly, it is possible to use various methods as the method for the asynchronous secondary device 200 R (e.g. backup module 242 R, FIG. 7) to judge whether or not the completion notification of all the asynchronous copy requests of a certain generation have been issued. For example, the same as with the synchronous secondary device 200 L, the asynchronous secondary device 200 R may also receive copy requests to which the generation number GNo (FIG. 12) has been added. In this way, the asynchronous secondary device 200 R is able to judge in the same way as the synchronous secondary device 200 L. Also, in this case, the history creation module 238 R is able to set the generation of the third bitmap BMC by referencing the generation number GNo added to the copy request.

Variation Example 6

For each of the embodiments noted above, the asynchronous secondary device 200 R, the same as with the synchronous secondary device 200 L, may also receive copy requests to which are added a sequence number SNo (FIG. 12). Furthermore, it is also possible for the asynchronous secondary device 200 R to execute the write processing according to requests in the sequence of this sequence number SNo. In this way, the asynchronous copy destination volume VOL 1 -R always has host consistency. Therefore, it is possible to omit the backup volume VOL 2 -R and the backup module 242 R. However, as with each of the embodiments noted above, the asynchronous secondary device 200 R may write the data of each request to the volume VOL 1 -R in the sequence in which the copy requests are received. In this way, it is possible to use a simple constitution of the asynchronous secondary device 200 R. These are also the same for the synchronous secondary device 200 L.

Variation Example 7

For each of the embodiments noted above, the secondary copy module 240 L, the same as with the copy module 232 , preferably has the instruction module and the access module for executing data copying. In this way, it is possible to reduce the effort by the user required for setting the copy destination storage area of the secondary copy process.

Variation Example 8

For each of the embodiments noted above, it is also possible to replace with hardware the parts of the constitution realized using software, and conversely, to replace with hardware the parts of the constitution realized using hardware. For example, the functions of the read/write module 235 (FIG. 2) may be realized by a hardware circuit having logic circuits.

Although the present invention has been described and illustrated in detail, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the spirit and scope of the present invention being limited only by the terms of the appended claims.